In the current study, tumor location was found to be the only variable significantly associated with the development of LMD after postoperative SRT. The crude incidence of 14% with an actuarial incidence of 18% after two years is in line with the incidence between 10 and 20% described in the literature.(5, 16-20)
Our finding that LMD is more common after treatment of cerebellar metastases compared to cerebral metastases is consistent with other studies.(21-25) The close proximity of cerebellar BMs to the large CSF spaces in the posterior fossa, like the cisterna magna, might be a reason for the increased risk of LMD in these patients.(24, 25) These CSF spaces could act as reservoirs for tumor cell spillage during surgery, which may increase the risk of LMD. The comparatively small subarachnoid space which contains CSF surrounds superficial cerebral metastases, so intraoperative tumor spill may be less likely to occur here. Close proximity of BMs to any CSF space could be theorized to lead to an increased risk of LMD.(26) In our study, patients with BMs close to CSF spaces were found to be at increased risk of LMD, although this effect was not statistically significant (HR 4.89, 95% CI 0.86-27.9, P 0.074).
Neurosurgery with postoperative SRT has been associated with an increased risk of development of LMD compared to SRT only.(4, 5) Limiting intraoperative tumor dissemination remains challenging due to a limited number of known surgery-specific risk factors. One previously reported risk factor is tumor spillage after a piecemeal resection method rather than en bloc.(15, 27) Likewise, intraoperative ventricle violation has been associated with increased risk of LMD.(28) Previous studies have reported several other risk factors for developing LMD, including multiple BMs at baseline, younger age, large preoperative tumor size, presence of extracranial metastases, and breast cancer as the primary tumor location.(4, 5, 16, 21, 22, 25-27, 29-33) Our study found an association between some of these factors and increased risk of LMD, but only the tumor location was statistically significant. This may be due to a smaller study cohort with less total events compared to other studies or due to the limited number of patients from specific subgroups such as breast cancer patients. A larger cohort could potentially have shown more factors significantly associated with LMD.
This study has several limitations which should be considered when interpreting the results. The results of this retrospective study need to be confirmed by prospective studies to reduce the effect of selection bias. Additionally, the diagnosis of LMD was based on radiological findings in the majority of cases rather than confirmation from CSF cytology. The radiological diagnosis of LMD can be challenging, leading to interobserver variability in interpretation between radiologists and potentially inconsistent findings. CSF cytology can likewise produce false-negative results, which may lead to underestimation of the true incidence of LMD. Despite these limitations in diagnostic methods, our observed incidence rates are comparable to those reported in other literature.(5, 16-20) Furthermore, the groups were unevenly distributed in terms of BM location, with only 27% of BMs located in the cerebellum. This explains the relatively wide range of the 95% confidence interval in this group (Table 2). Lastly, LMD may have been missed in undiagnosed patients who die outside a hospital setting. Despite the aforementioned limitations, the study possesses several strengths, including a relatively homogenous patient population, a relatively long follow-up interval, consistent treatment approaches, and regular MRI follow-up. The findings of the study are consistent with those of previous research.(5, 16-20)
After diagnosis of LMD, treatment options are primarily decided by general patient condition as well as the availability of targeted systemic therapies with adequate blood-brain barrier penetration (such as ALK- and BRAF-inhibitors). Systemic treatments when available can be considered based on primary tumor characteristics and previous treatments. Chemotherapy can be administrated intrathecally to bypass the blood-brain barrier, but is generally no longer advised in the Netherlands after a randomized study found no survival benefit after adding intrathecal methotrexate.(34, 35) In the latest European Society for Medical Oncology (ESMO) guideline, SRT is recommended in order to treat symptomatic nodular sites of LMD in the brain and spine, while whole brain radiotherapy (WBRT) is only recommended for patients with extensive nodular or symptomatic linear LMD.(36) Neurosurgery is only recommended to relieve symptoms in some patients with symptomatic hydrocephalus, but this treatment comes with a risk of complications and an additional treatment burden.(37) In the current study, survival time after LMD diagnosis was generally limited, especially in patients where best supportive care was started: no patient in this group lived longer than 25 days after diagnosis.
Since patients with cerebellar metastases are at increased risk of developing LMD, that group is likely to benefit most from alternative treatment strategies aimed at mitigating this risk, such as pre-operative SRT. Nevertheless, patients with cerebral metastases were three times more prevalent than cerebellar metastases in this study, accounting for over half of LMD cases. Therefore, studies on treatments like pre-operative SRT ought to include patients with cerebral metastases as well.
The relatively high prevalence of LMD highlights the importance of finding alternative treatment strategies to prevent the detrimental effects of LMD. It is hypothesized that preoperative SRT enables the sterilization of tumor cells before the intraoperative spillage of malignant cells into the CSF can occur.(38-40) Retrospective studies suggest that preoperative SRT is associated with a lower risk of LMD compared to postoperative SRT, while there is no difference compared to WBRT.(7, 41, 42) The largest comparative studies were performed by Patel et al. A retrospective study from this group (n = 102) showed no difference in overall survival, local/distant recurrence, and LMD rates between preoperative SRT and postoperative WBRT.(41) In another study (n = 180), the same group found higher rates of LMD (17% versus 3%, P = 0.01) two years after postoperative SRT compared to preoperative SRT.(42) Preoperative SRT additionally decreases dose exposure in healthy brain tissue due to a more clearly defined PTV compared to the postoperative situation.(43) Likely due to the smaller PTV, the incidence of radionecrosis is lower after preoperative SRT.(40, 41, 44) Additionally, there are no increased risks of complications such as delayed wound healing following preoperative SRT and surgery.(6, 7, 45) The main challenge of implementing preoperative SRT appears to be the logistical challenge of adopting it into current treatment workflows, especially for symptomatic patients who require early neurosurgery for large BMs.(44) Optimal timing and fractionation schedules are yet to be determined. The majority of published studies generally reported surgery taking place within 48 hours after SRT.(46) Radiation doses of 15-30Gy are typically delivered in 1-5 fractions.(44) Currently, several prospective studies comparing pre- and postoperative SRT are ongoing in Europe and North America, including four randomized trials (ClinicalTrials.gov identifiers: NCT03741673, NCT05124236, NCT04474925, NCT03750227).