Is local radiotherapy a viable option for patients with an opening of the ventricles during surgical resection of brain metastases?

Local therapies, such as surgery and radiosurgery, have become the standard of care in patients with a limited number of brain metastases as they combine excellent outcomes with a reduction of neurological side effects compared to whole brain radiotherapy (WBRT) [1‐3]. Surgery of solitary brain metastases has been shown not only to reduce seizures, symptoms of intracranial pressure and neurological deficits but also to improve overall survival (OS) in patients with good performance status and controlled systemic disease compared to WBRT [4].

In order to avoid cognitive decline, adjuvant radiotherapy to the resection cavity is also shifting towards a local approach [2, 3, 5‐7]. Local radiotherapy to the resection bed can be administered by radiosurgery with single doses of 15–22 Gy or hypofractionated stereotactic radiotherapy (HFSRT) [2, 3, 6, 8‐11]. Previously, our group has demonstrated excellent local control after HFSRT to the resection cavity of approximately 80% [6, 8]. Furthermore, we showed that distant intracranial recurrences and even most local failures can be salvaged successfully without an excessive risk for radionecrosis [12].

The reduction of neurological side effects of local radiotherapy comes at a price. Both, surgery and local radiotherapy entail higher rates of leptomeningeal disease (LMD) and distant failure compared to WBRT [13]. In the case of surgery without adjuvant radiotherapy local control is reduced as well [14]. The risk of LMD development after surgical resection of brain metastases occurred particularly in patients with posterior fossa metastases and could be reduced significantly by changing the resection technique to “en-block” resection as compared to “piecemeal” [4, 15‐17]. Different factors may contribute to an increased LMD manifestation including the tumors natural biology and local radiotherapy leading to higher incidences of distant tumor progression resulting in LMD.

The contamination of cerebrospinal fluid during the course of surgery, nevertheless, should not be ignored as a potential trigger mechanism leading to an increase in LMD after surgical resection [11, 15, 16]. In periventricular metastases surgical resection can lead to an opening of the ventricles. Subsequently it carries a potential risk of cerebrospinal tumor cell dissemination [17‐21]. The aim of this study was to assess whether local HFSRT of the resection cavity is a viable option in patients with resected brain metastases in which VO was necessary during surgery.

All patients were treated in accordance with the Declaration of Helsinki. A written informed consent in the use of scientific data was obtained by all patients. This study was approved by the Ethics Committee of the Technical University of Munich.

We analyzed the outcome of 119 patients with 123 resection cavities after surgical resection and local radiotherapy to the resection cavity in order to evaluate the efficacy of local radiotherapy in patients with an opening of the ventricles during surgery. No difference in local control, distant brain control, development of LMD or OS was noted when comparing patients with and without VO. The only difference observed between the two groups was a significantly increased rate of distant intraventricular metastases in patients with VO.

Radiosurgery and HFSRT are both practicable concepts to achieve local tumor control after surgical resection of brain metastases. A large randomized study demonstrated improved cognitive functioning for radiosurgery vs. whole brain radiotherapy yet at the price of reduced distant and local control [2]. HFSRT may overcome the disadvantage of the higher local failure rate as fractionation permits the use of increased safety margins to compensate for infiltrative growth and uncertainties in target volume due to volumetric changes of the cavity [22‐24]. Multiple retrospective studies have demonstrated outstanding local control rates for resection cavities treated with HFSRT [6, 8, 10, 12]. The one year local control rate of a cohort from our clinic and a combined analysis with another institution applying a similar radiation scheme of 30/35 Gy in 5 Gy fractions was 88% and 80.5% [6, 8]. This is noticeably superior to the one year local control rates of 60.5% and 72% reported by Brown et al. and Mahajan et al. with radiosurgery and comparable to the WBRT control group [2, 3]. Other groups have shown similar local control rates with different fractionation schemes [26, 27]. Furthermore, the safety profile of HFSRT seems to be favorable compared to radiosurgery, particularly for large cavities [9, 28]. In comparison to WBRT, on the other hand, the risk of radionecrosis seems to be higher for HFSRT [2, 6, 11, 29, 30]. A potential downside to HFSRT and radiosurgery is the increased risk of leptomeningeal spread and lower distant brain control [2, 26, 30‐32]. While localized distant progression can be easily salvaged after local radiotherapy to the resection cavity, LMD is a severe complication that considerably diminishes the patients’ survival [12, 33].

Surgical resection of brain metastases as well carries a risk of intracerebral tumor cell dissemination. LMD rates as high as 36% have been reported particularly in patients with piece meal resection of posterior fossa tumors [15‐17]. An opening of the ventricles during surgery creates a connection between the resection cavity, the location with the highest risk for recurrence and the cerebrospinal fluid. A dissemination of tumor cells through the CSF could be a potential consequence. Therefore, the aim of this study was to assess whether local HFSRT of the resection cavity is a viable option in patients with resected brain metastases in which VO was necessary during surgery.

Neither distant brain control nor LMD development was significantly associated with an opening of the ventricles. Therefore, we assume local radiotherapy to be sufficient in these cases. However, even if numbers are not statistically different, it seems that the 2-year intracranial control in the VO subgroup is arithmetically lower. The total number of events is clearly limited, possibly disguising differences.

The development of distant intraventricular metastases was significantly increased in patients with an opening of the ventricles pointing towards a potential dissemination of individual tumor cells without the capacity to induce disseminated LMD. Since adequate therapeutic strategies for intraventricular metastases such as resection or local radiotherapy exist and neither OS nor distant control were affected by the development of distant intraventricular metastases, no change in target volume definition seems necessary due to our findings [34‐36]. An alternative explanation for the higher distant intraventricular failure rate is the larger preoperative metastatic diameter of metastases with VO, as larger metastases are associated with increased distant failure rates [28]. Further thought should be given to target volume definition in this particular situation. Recently, contouring guidelines for stereotactic radiotherapy to resection cavities have been established recommending an additional margin of 10 mm for metastases with dural contact and up to 5 mm for metastases with venous sinus contact [37]. Whether an additional ependymal margin is of value cannot be answered by this study, since our standard contouring concepts add 4–5 mm margins as we apply fractionated radiotherapy. A potential alternative is preoperative radiotherapy which permits an easier definition of the target volume [38]. However, preoperative radiotherapy also is accompanied by potential pitfalls such as the absence of histopathological specimen at the time of radiotherapy and a possible increase in wound complications [41].

A comparable study has been conducted by Adeberg et al. for patients with glioblastoma that experienced an opening of the ventricles during surgical resection. Similarly to our results, no increased incidences of distant brain recurrence were noted in the study [39]. However, a number of studies demonstrated higher rates of LMD in high grade glioma patients after VO [20, 42, 43]. Moreover, the dissemination pathways of glioblastoma and metastases are likely different.

With roughly 6% the overall incidence of LMD was relatively low. LMD occurred more frequently in certain histologies such as NSCLC, melanoma and breast cancer, which is in line with findings by other groups [40]. The incidence of LMD was not significantly higher in patients with infratentorial metastases which might be due to improved surgical techniques.

Limitations to our study include the low incidence of VO and subsequent small patient and event numbers. Furthermore, we lack information on the systemic therapies applied. Therefore, we cannot affirm that the distribution of systemic therapies such as immunotherapies, that might influence the outcome, including the risk of intracranial dissemination, was equal between the groups.

The fact that none of the patients with VO developed LMD, nevertheless, underlines the hypothesis that local radiotherapy is viable in patients with VO during surgical resection of brain metastases. However, further studies should be conducted to clarify the subject.

VO during neurosurgical resection did not affect the outcome after HFSRT of the resection cavity in patients with brain metastases. Particularly, the incidence of LMD was not increased in patients receiving local HFSRT after VO. HFSRT can therefore be offered independently of VO as a local treatment of tumor bed after resection of brain metastases.