1. Introduction
Brain metastases (BM) are a common and devastating complication in the clinical course of systemic malignancies [
1,
2]. These secondary brain tumors constitute the most frequent malignant neoplasms of the central nervous system with an increasing incidence rate [
3]. Neurosurgical tumor resection is an important treatment option in the multimodal management of BM. The aim of surgery represents the complete and safe removal of detectable BM tissue. Nevertheless, incomplete resection of BM is not uncommon in the routine neurosurgical practice, resulting in high rates of local recurrence, even despite initiation of postoperative adjuvant therapies [
4,
5,
6].
One major reason for such incomplete tumor resections might be insufficient detection of BM tissue during neurosurgical resection. Indeed, an unexpected residual tumor after surgery was identified by postoperative magnetic resonance imaging (MRI) in approximately 25% of BM [
7,
8]. Additionally, the peritumoral brain tissue of BM is assumed to be an important site for residual tumor tissue. In this sense, recent studies observed a high rate of BM with infiltrative growth into the peritumoral brain tissue although these tumors were initially considered as well-demarcated neoplasms [
7,
9,
10]. Typically, this infiltrative behavior of BM consists either of a growth along pre-existing blood vessels in a so-called “vascular co-option” growth pattern or a diffuse “glioma-like” single cell infiltration of peritumoral brain tissue [
9]. Due to evolving targeted therapies, angiogenesis of BM attracted more attention in the neurooncological field in the last few years [
11,
12]. However, the presence of angiogenesis in the peritumoral brain tissue and its potential impact on progression/recurrence of BM has not been clarified so far.
An innovative approach for improved intraoperative visualization of malignant brain tumor tissue is the application of 5-aminolevulinic acid (5-ALA) fluorescence [
13,
14]. Primarily, this technique was used for fluorescence-guided resections of high-grade gliomas [
15]. Lately, visible 5-ALA fluorescence was observed in BM in the first patient series as well [
16,
17]. Although the majority of BM displayed a visible but heterogeneous 5-ALA fluorescence pattern, the impact of 5-ALA in BM still remains controversial [
16,
17]. Interestingly, visible 5-ALA fluorescence was recently also detected in the peritumoral brain tissue after BM resection in a subgroup of patients [
16,
17]. However, to date, the significance of this visible 5-ALA fluorescence has not been determined. If visible 5-ALA fluorescence could be used to identify tumor infiltration and/or angiogenesis in the peritumoral brain tissue, the technique might be useful for guidance of individualized perioperative treatment concepts.
The aim of this study was thus to analyze the histopathological correlate of visible 5-ALA fluorescence in the peritumoral brain tissue of BM and its influence on patient prognosis. For this purpose, we investigated specific histopathological parameters such as tumor infiltration and angiogenesis in a large series of tissue samples from fluorescing and non-fluorescing peritumoral tissue collected during resection of BM.
3. Discussion
Tumor cell infiltration of the peritumoral brain tissue is a common finding in BM that might result in local recurrence and thus worse patient prognosis [
9,
18]. However, intraoperative visualization of such tumor cell infiltration is still challenging. Two decades ago, the 5-ALA fluorescence technique was introduced as an innovative method for improved visualization of tumor tissue during surgery of high-grade gliomas and is nowadays applied to safely increase the EOR at many neurosurgical centers worldwide [
13,
14,
15]. In the largest series to date, we found visible 5-ALA fluorescence also in 66% of 157 BM [
17]. Interestingly, we observed visible 5-ALA fluorescence also in the peritumoral brain tissue after assumed complete BM resection in a subgroup of patients [
17]. Thus, we speculated that 5-ALA fluorescence might visualize tumor cell infiltration in adjacent brain tissue of BM and collected 88 fluorescing and non-fluorescing samples from this region for histopathological analysis.
In this study, we found visible 5-ALA fluorescence in the peritumoral brain tissue during BM surgery in 76% of cases. Surprisingly, preoperative chemotherapy of the primary tumor and local radiation treatment preceding the resection of BM did not show a significant correlation with 5-ALA fluorescence, neither in the BM nor in the peritumoral brain tissue. We found tumor cell infiltration of the peritumoral brain tissue in 31% of BM. According to the data, we did not find a significant correlation between the 5-ALA fluorescence status of the peritumoral brain tissue and tumor cell infiltration. However, this might be influenced by a selection bias, since the number of samples taken from the peritumoral brain tissue was low (median 1; range: 1–4) and identification of tumor cell infiltration could simply be missed. Nonetheless, similar to our study, Kamp et al. found tumor cells in only 33% of specimens from the resection cavity of BM with visible 5-ALA fluorescence [
16]. It is of note that a large portion of cases demonstrated visible 5-ALA fluorescence in the peritumoral brain tissue, albeit the BM itself did not show any visible 5-ALA fluorescence. Therefore, we assume that visible 5-ALA fluorescence in the peritumoral brain tissue of BM is not primarily caused by tumor cell infiltration. The lack of correlation between visible 5-ALA fluorescence in the peritumoral brain tissue and presence of tumor tissue might be induced by diffuse leakage of protoporphyrin IX, the fluorescing substrate within the 5-ALA metabolism, into the adjacent brain tissue. Similarly, Utsuki et al. discussed the “false positive” effect of 5-ALA fluorescence in malignant brain tumors [
19]. Namely, the authors detected immune-cell infiltration and reactive astrocytes in fluorescent areas of peritumoral edema rather than tumor cell infiltration [
19]. Additionally, Siam et al. found unique glial activation, a “glial-defense system”, at the tumor-brain-interface [
18]. We assume, that these important co-factors might additionally influence visible 5-ALA fluorescence in the peritumoral brain tissue, which challenges the current understanding of the mechanism of 5-ALA fluorescence. Therefore, in the scope of a future study, we intend to address these crucial points and investigate immune-cell infiltration in fluorescing peritumoral brain tissue of BM. Nevertheless, a recent study found a significant correlation between 5-ALA fluorescence in the peritumoral brain tissue and tumor cell infiltration [
20]. However, the authors relativized their findings since visible fluorescence was also found in peritumoral brain tissue with only very scarce tumor cell infiltration [
20].
Further, in our study, we did not observe a significant association of visible 5-ALA fluorescence in the peritumoral brain tissue or tumor cell infiltration and time to progression/recurrence or one-year survival. In the literature, only reports correlating the 5-ALA fluorescence status of BM tumor bulk with prognostic factors are available. In this sense, Kamp et al. assumed in two of their studies that the 5-ALA fluorescence status in the tumor tissue of BM might be considered as a prognostic marker with a more aggressive behavior in BM along with absence of fluorescence [
21,
22]. Siam et al. described a significantly lower survival rate in patients with tumor infiltrated peritumoral brain tissue compared to patients without tumor infiltration [
18]. However, the conflicting findings between the study of Siam et al. [
18] and our study might be related to tissue sampling differences. In contrast to our study, Siam et al. [
18]. included only BM with non-eloquent localization; thus, a higher number of samples from the peritumoral brain tissue (median: three samples per patient) could be collected compared to our study including also BM with eloquent localization (median: one sample per patient). By collecting a higher number of samples per BM, the likelihood was increased to detect tumor cell infiltration in at least one of the collected specimens [
18]. We suspect a regional heterogeneity of tumor cell infiltration in brain tissue adjacent to BM, which has to be further investigated in future studies, especially in view of patient prognosis.
Surprisingly, angiogenesis was significantly related to the 5-ALA fluorescence status in our study since it was only found in fluorescing samples of peritumoral brain tissue. In line, a previous study of Roberts et al. in glioblastoma patients also suggested that approximately 4% of the specimens with visible 5-ALA fluorescence were either fully necrotic or showed abnormal, prominent vasculature [
23]. It is of note that we also observed a significantly shorter time to local progression/recurrence (2 vs. 22 months) and lower one-year survival rate (23% vs. 52% one-year survival) in BM with angiogenesis in the peritumoral brain tissue compared to BM without angiogenesis in the adjacent brain tissue. To our best knowledge, the association of 5-ALA fluorescence in the peritumoral brain tissue of BM and presence of angiogenesis as well as the negative impact of angiogenesis on the time to progression/recurrence and one-year survival has not been described so far. Biologically, the observed angiogenesis might represent tumor induced preparation of the peritumoral brain tissue to allow vascular co-optive growth [
9,
24]. Indeed, the vascular co-optive growth pattern was frequently observed in autopsy specimens of BM [
9,
24]. Since visible 5-ALA fluorescence significantly correlated with angiogenesis, the intraoperative 5-ALA fluorescence status of the peritumoral brain tissue in BM might guide individualized perioperative treatment concepts. Moreover, due to the fact that angiogenesis was never found in non-fluorescing brain tissue in our study, lack of 5-ALA fluorescence in the peritumoral brain tissue might be an intraoperative indicator for absence of angiogenetic features. In contrast, angiogenesis might be present in case of visible 5-ALA fluorescence and, thus, a supramaximal resection with additional safe removal of a “safety margin” of 5 mm from the peritumoral brain tissue could be considered especially in BM with non-eloquent localization to improve the patient prognosis [
25]. In this sense, Yoo et al. demonstrated an improved local tumor control and a significantly improved two-year survival in comparison to complete resection of the BM alone (63% vs. 29% two-year survival) [
25]. In case of eloquent BM localization and non-resectable peritumoral brain tissue with biopsy confirmed or suspicion of angiogenesis a modification of the postoperative radiotherapy plan with extension of the radiation field around the resection cavity might be considered. However, angiogenesis was absent in 48/61 (79%) fluorescent samples of the peritumoral brain tissue. This might also explain the non-significant correlation between fluorescence in peritumoral brain tissue and one-year survival. Nevertheless, postoperative treatment such as local radiation therapy and WBRT did not show a significant influence on time to local progression/recurrence, whereas angiogenesis remained an independent factor even in multivariate analysis. According to univariate and multivariate analyses, postoperative treatment (local radiation therapy and WBRT) as well as adjuvant chemotherapy did not have any significant effect on one-year survival compared to angiogenesis and GPA class, which remained independent factors in multivariate analysis. The association of the standard prognostic score GPA (graded prognostic assessments) underscores that we investigated a regular BM cohort with a certain range of prognostic parameters. Altogether, future studies with a larger number of BM are needed to confirm our preliminary results and extensively investigate the significance of angiogenesis/fluorescence in the peritumoral brain tissue of BM.
Although our study is among the first to investigate the association of 5-ALA fluorescence with tumor cell infiltration and angiogenesis in BM, we have to interpret the results in the light of some limitations. In this sense, we were able to include only a fraction of patients treated with resection of BM as collection of tissue samples of the peritumoral brain tissue was not feasible in all patients especially due to tumor localization in highly eloquent areas. We analyzed one full section of each BM and specimens of peritumoral brain tissue. As serial sectioning was not performed, different patterns of infiltration and angiogenesis throughout the tumor sample cannot entirely be ruled out. In our view, the volume of the biopsies (approximately 80–125 mm3) of the single section is sufficient to be considered as representative of the total volume of peritumoral brain tissue. However, it cannot definitely be excluded that specific pathological features leading to 5-ALA fluorescence (i.e., angiogenesis, tumor cell infiltration, or other) are missed in this single section. Moreover, the number of samples taken from the peritumoral brain tissue was limited (median 1; range: 1–4) and identification of tumor cell infiltration could be missed. Consequently, we have to interpret out results with caution. Nevertheless, although the limited number of samples certainly might cause a bias, the current study is, to our best knowledge, the first to systemically explore angiogenesis in the peritumoral area of resected BM. Additionally, one must consider that, in certain cases, peritumoral brain tissue of BM includes functional brain tissue and, thus, cannot be biopsied without harming the patient. Another limitation is that the rate of samples from peritumoral brain tissue in our study might be lower in cases with a poor margin. Therefore, our strategy of tissue collection only from safe sampling localizations might have systematically biased against cases with a poor margin. Thus, we suggest that our findings regarding angiogenesis and patient prognosis should be regarded carefully. Additionally, our patient cohort is quite small; therefore we assume that analysis including a larger patient cohort might show differing results. Even more, if more extensive tissue sampling and serial sectioning is performed. Therefore, we suggest, that our observed preliminary findings and certainly the impact on the future treatment of BM patients need to be investigated in larger clinical trials.
4. Materials and Methods
Adult patients (≥18 years) with neurosurgical resection of BM after preoperative 5-ALA administration at the Department of Neurosurgery, Medical University Vienna, were included. The utilization of 5-ALA for the resection of BM was determined by the individual treating neurosurgeon. 5-ALA was not administered in patients with BM who required urgent surgery due to acute mass effect and/or hydrocephalus. At our institution, a mandatory precondition for elective resection of BM is the presence of a stable primary disease confirmed by the attending oncologist. Additionally, patients with known contraindications to 5-ALA were not included in this study. Patients were only included after giving informed consent to the participation in the present study. Furthermore, only BM were included in this study after histopathological confirmation. Brain tumors other than BM were excluded. Finally, only BM with available tissue samples from the peritumoral brain tissue with known 5-ALA fluorescence status were included. Our study was approved by the local ethics committee of the Medical University Vienna (EK: 419/2009; Amendment).
4.1. Preoperative Imaging and Clinical Data
The pattern of contrast enhancement (CE) on preoperative MRI was classified as “cystic”, “solid” and “heterogeneous” by an experienced neuroradiologist (J.F.). Quantification of the volume of the tumor and peritumoral edema was determined by volumetric measurements with a specialized software (Brainlab Elements SmartBrush, Version 2.6, Brainlab AG). Moreover, the localization of the resected BM as well as the number of BM (single or multiple) at time of surgery were determined. The Graded Prognostic Assessment (GPA) class as a prognostic index for patients with BM and clinical data were retrieved from the hospital archive system (AKIM) and the Vienna Brain Metastasis Registry.
4.2. 5-ALA Management and Neurosurgical Resection of BM
A standard dosage of 5-ALA (20 mg/kg bodyweight) was administered orally approximately three hours prior to anesthesia. For visualization of potential 5-ALA fluorescence a modified surgical microscope (NC4/Pentero, Carl Zeiss Surgical GmbH, Oberkochen, Germany) was applied. All patients underwent neurosurgical resection of BM with assistance of a neuronavigation system for intraoperative guidance. For maximum safe BM resection in eloquent brain areas, navigation with diffusion tensor imaging/functional MRI and/or intraoperative monitoring including cortical/subcortical mapping and stimulation was used to minimize the risk of a new postoperative neurological deficit. During surgery, the visible fluorescence status (visible or no fluorescence), fluorescence level (strong, vague or negative) and fluorescence homogeneity (homogeneous or heterogeneous) of each resected BM was determined by the performing neurosurgeon as described previously [
17]. Generally, tissue sampling from the peritumoral brain tissue was conducted after assumed complete resection of the BM. For tissue sampling we usually applied a biopsy forceps with a diameter of the tip of 5 mm (Aesculap
® microform, FD216R). Tissue samples from the peritumoral brain tissue were safely collected only in localizations without the risk of causing postoperative neurological deterioration. The site of tissue collection was primarily based on the fluorescence status of the peritumoral brain tissue and the safety of tissue collection. Thus, tissue sampling did not usually depend on macroscopic factors such as suspicious vessels in the peritumoral brain tissue or suspected infiltrated white matter. In detail, the resection cavity was checked for potential 5-ALA fluorescence in the peritumoral brain tissue after assumed complete BM removal. In case of visible fluorescence, tissue samples were safely collected from fluorescing regions. In case of absence of visible fluorescence, tissue sampling from the peritumoral brain tissue was performed from localizations with maximum possible distance from eloquent brain areas. The 5-ALA fluorescence status (visible or no fluorescence) was documented for each tissue sample from the peritumoral brain tissue. Due to potential occurrence of significant brain-shift following tumor removal, we did not systematically use image guidance with neuronavigation for precise localization of the peritumoral brain tissue. Patients were finally included in the current study if at least one tissue sample of the peritumoral brain tissue could be safely collected during BM surgery.
4.3. Histopathological Assessment
A board-certified neuropathologist verified the diagnosis of BM. Additionally, all collected tissue samples from the peritumoral brain tissue were further investigated. For this purpose, presence of tumor cells, different infiltration patterns and angiogenesis were evaluated on hematoxylin and eosin stains on one section per sample. The infiltrative tumor growth pattern was classified as either diffuse single cell infiltration or vascular co-option as described previously [
9]. Moreover, presence of angiogenesis in the peritumoral brain tissue was defined as vessels presenting with multilayered endothelia as previously determined [
26]. As outlined in
Figure 4A–F vessels with a single endothelial layer were defined as “no angiogenesis” while presence of a multiple endothelial cell layer as well as diversity in vessel configuration was defined as angiogenesis. If more than one sample was available per tumor, one representative specimen was selected to indicate the status of tumor cell infiltration or angiogenesis of each BM. Thus, tumor cell infiltration or angiogenesis was considered to be present in the peritumoral brain tissue, if at least one sample showed histopathological features of tumor tissue or angiogenesis. The different growth patterns and angiogenesis are illustrated in
Figure 4.
4.4. Postoperative Course
After surgery, patients were protected from strong light sources to avoid potential 5-ALA related phototoxic side effects for a minimum time period of 24 h. Routinely, a postoperative computerized tomography of the brain was performed on the first postoperative day to exclude intracranial hematoma or ischemia. In the last years, postoperative MRI preferentially within 72 h after surgery was implemented into our standards of perioperative patient management in BM to assess their EOR. The postoperative neurological course at discharge was compared to the preoperative neurological status and classified as unchanged, improved or deteriorated. In patients with postoperative neurological impairment, the symptoms were reevaluated for potential improvement in the 3 months follow-up visit. In each patient, postoperative treatment was individually determined by an interdisciplinary tumor board. Radiotherapy was performed according to standard protocols used by our Department of Radiooncology. Therefore, supramarginal radiotherapy of the resection cavity was performed with a total of 35 Gy and 5 Gy in each fraction. SRS was either performed with lineac or gamma knife radiation. Hippocampal sparing WBRT was not applied in the present series. For WBRT, 10 fractions of 3 Gy (total 30 Gy) were applied. Data on time to local progression of a residual tumor, local recurrence after complete resection in the region of the surgically treated BM or distant recurrence in the brain at another site than the surgically removed BM were retrieved from regular follow-up MRI assessed by the experienced neuroradiologist (J.F.) whenever available. The data on one-year survival were obtained from the Vienna Brain Metastasis Registry.
4.5. Statistical Analyses
All statistical analyses were performed with SPSS 26.0 software (SPSS Inc., Chicago, IL, USA). Inferential analyses were used as appropriate by the use of Chi
2-tests, Fisher’s-exact test and the Mann–Whitney U test for assessment of differences in 5-ALA fluorescence status of BM or peritumoral brain tissue with the volume of tumor/edema, primary tumor type and CE on preoperative MRI. Furthermore, differences between 5-ALA fluorescence in the collected tissue samples and tumor cell infiltration, growth pattern and angiogenesis were investigated. Due to the exploratory and hypothesis generating design of the present study no adjustment for multiple testing was applied [
20]. The primary endpoint in our patient cohort was determined as the one-year survival time, which was defined as the time from diagnosis of BM to death within the first 12 months after surgery. Kaplan–Meier curves were used to estimate the one-year survival and the log-rank test was applied to investigate group differences. For statistical analyses we grouped the different postoperative treatments into (1) no postoperative treatment, (2) local radiation treatment (SRS and radiotherapy of resection cavity) and (3) WBRT. Further, univariate analyses were performed according to a Cox regression hazards model. Variables with significant univariate results were entered into a multivariable Cox regression hazards model. A
p-value < 0.05 was considered as statistically significant.