Subgroup analyses

Apart from fractionation schemes, there exist other parameters that might influence LC after RC irradiation. To evaluate possible further influencing factors, the PubMed database (https://pubmed.ncbi.nlm.nih.gov) was searched for the combined criteria "brain metastases", "resection cavity irradiation" and "local control". The literature search yielded 137 articles. After abstract review, 13 articles remained that appeared suitable to deliver relevant information, all retrospective in nature. There is no parameter that has consistently been demonstrated as a significant influencing factor on LC. However, those factors that showed a significant influence of LC in at least one dataset have been included as stratification factors for subgroup analyses. Subgroup analyses will be carried out using logistic regression analysis. The influencing factors identified are listed below, together with the supporting literature:

Although a radioresistant histology such as melanoma, colorectal carcinoma or sarcoma usually does not necessarily appear to have a significant impact on LC [22], Soliman et al. [23] detected a significant association between radioresistant primary and reduced LC.

Preoperative tumor size has also been described as a significant factor influencing local recurrence [24, 25]. A uniform cut-off value was not observed, therefore the categories "resected for size" (according to our in-house standard operating procedures approx. 3 cm in diameter) and "resected for clinical complaints" were defined for this subgroup analysis.

As shown by El Shafie et al. [26], incomplete resection may significantly lower LC rates.

The size of the resection cavity or the PTV volume has been demonstrated to significantly influence LC in several data sets [7, 15, 24–28], with bigger RC yielding poorer LC rates. Although absolute cut-off values differ, they are close to the median resection cavity volume / PTV volume in the respective data sets. In the SATURNUS trial, the size of the PTV is also dependent on the allocation to the treatment arm and positioning method. Therefore, the extent of the RC is the more appropriate stratification factor for this subgroup analysis.

As shown by Eitz et al. [15] > 1 metastasis present can have a significantly negative impact on LC.

In the multi-institutional analysis by Eitz et al. [15], an uncontrolled primary tumor yielded poorer LC rates.

There are a couple of analyses [24, 28, 29] showing a significant impact of delivered dose on LC. As 48 Gy BED10 has been demonstrated as cut-off value in two independent data sets [28, 29], this threshold is also used in the subgroup analysis.

Regarding the primary endpoint (LC of resected metastases) as well as LC of non-resected metastases, it will be analyzed exploratory whether an association between LC rate and prescription IDL as well as Dmax, Dmean and Dmin of the PTV can be observed. For the secondary endpoints LRC, OS and development of radionecrosis and pseudoprogression, possible influencing factors will be analyzed exploratively.