Nodular leptomeningeal disease after surgery for a brain metastasis—should we be concerned?

See the article by Prabhu et al. in this issue, pp. 1049–1059.

The approach to treating brain metastasis has evolved over the past few decades. Before the 1990s, patients were treated with palliative intent with whole brain radiotherapy (WBRT), if at all, and often further systemic treatment was halted. Therapeutic nihilism was reduced when a prospective, randomized clinical trial showed a survival advantage for surgery of a single brain metastasis plus WBRT versus WBRT alone and oncologists began to recognize a role for active treatment of localized brain involvement.¹ Subsequently, the advent of stereotactic radiosurgery (SRS), which permitted noninvasive treatment of brain metastases, facilitated focal treatment of even multiple metastases, and even provided a survival benefit for some patients, led to a paradigm shift in oncology.²

Yet, some established paradigms take time to change. Until recently, often patients with brain metastases were treated with WBRT, and subject to the attendant risk of neurocognitive decline. There were attempts to eliminate the use of WBRT, either as an adjuvant to surgery or SRS. A prospective, randomized trial evaluating WBRT after surgery showed a local failure rate of nearly 50% when WBRT was withheld.³ Three prospective, randomized trials of WBRT after SRS consistently showed poorer local control when WBRT was withheld, but 2 of the 3 showed evidence of neurocognitive decline in patients receiving WBRT.^4–6 These results, when taken together, indicate that WBRT after either surgery or SRS improves local control, most dramatically for surgery, but at a cost of neurocognitive decline.

The desire to improve local control and minimize the neurocognitive consequences associated with adjuvant treatment led to the development of adjuvant SRS as a new strategy. A prospective trial that randomized patients undergoing surgery to adjuvant WBRT versus SRS showed no difference in overall survival, but the patients in the SRS arm had a slower rate of neurocognitive decline (albeit with a median 2-week difference).⁷ These results appeared to support the general use of adjuvant SRS in lieu of WBRT following resection.

Other reports raised concerns about the use of adjuvant SRS and its ability to achieve loco-regional control. In a prospective, randomized trial of adjuvant SRS versus observation after resection of a brain metastasis, the local control rate in the SRS group was 72% (compared with 82% in the WBRT group in the Patchell study³). Remarkably, local control for the observation group was only 45%.⁸ This result occurred despite the use of an “en bloc” resection in 73% of cases; for the other 27% of cases, an “en bloc” resection was not feasible. In the end, the type of surgery (“en bloc” vs piecemeal) did not appear to have an impact on the risk of local recurrence; a change in surgical technique did not change the local control outcome that had been reported 30 years earlier.

A more substantial concern raised by the use of adjuvant SRS was that patients were developing what is now recognized to be a novel form of leptomeningeal disease (LMD), termed “nodular” or “pachymeningeal” LMD (nLMD). Nodular LMD appears as tumor masses in the extra-axial space, typically near the site of a surgical cavity. The appearance and timing of nLMD is strongly suggestive that it arises from microscopic tumor spillage at the time of surgery or spread in the postoperative period prior to adjuvant treatment. Given the lack of improvement in local control associated with a change in surgical technique, a likely explanation for the current observation of nLMD is that it is a direct consequence of the change in practice from adjuvant WBRT to adjuvant SRS. That is, adjuvant WBRT treated both the cavity and the surrounding tissues, and hence any microscopic tumor remnants and spillage were effectively treated before they could grow to become new masses. Adjuvant SRS, on the other hand, is designed to treat only the resection cavity seen on the planning MRI with minimal or no margins; microscopic tumor cells in the extra-axial space or surgical access track may not be fully included in the SRS treatment field. The use of adjuvant SRS, therefore, appears to have unmasked the tumor cell spillage that inevitably occurs with surgery for brain metastases.

Questions have been raised as to the actual rate of the phenomenon of nLMD and its clinical impact. Estimates of the risk of nLMD have ranged from 5% to 31%.⁹ Interestingly, if the order of treatment is changed, and SRS is administered before surgery, the risk of nLMD has been reported to be as low as 3.2%.¹⁰ This “neoadjuvant” SRS approach may offer a solution to the problem of occult tumor spillage following surgery, but the question remains: Is the problem of nLMD impactful? In this issue of Neuro-Oncology, Prabhu and colleagues have carefully evaluated the occurrence of LMD after surgery for a brain metastasis and classified it as to whether it is spontaneous (“classical” LMD, or cLMD) versus surgically induced (nLMD).⁹ They identified 147 patients at 7 hospitals who developed some form of LMD and estimated the overall risk for LMD was about 21% of all cases treated with surgery and adjuvant SRS. The nLMD patients constituted nearly 60% of the cohort, and about half of them presented with new symptoms. Salvage therapy most often involved the use of WBRT. Patients with nLMD lived substantially longer than those with cLMD but only when salvage radiation was used. In a related study, Cagney et al had found that 72% of patients with nLMD died from neurological progression if not treated.¹¹ Hence, the occurrence of nLMD produces a need for salvage treatment with WBRT that, in turn, impacts on quality of life and neurocognitive function. The problem of nLMD is one that potentially can be avoided via a change in treatment sequencing; a prospective, randomized trial of neoadjuvant versus adjuvant SRS after surgery for a brain metastasis is the best way to answer the question definitively.