Brain metastases (BM) occur in up to 40% of all patients with solid tumors over the course of disease (1, 2). Patients suffering from lung carcinoma, both non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC), as well as breast cancer and malignant melanoma are most commonly affected (1–3). Due to a short median survival time of less than 12 months across nearly all primary sites and the often-limited efficacy of systemic therapy, clinical management of BMs can be exhausting and requires multidisciplinary expertise (1, 2). According to the 2021 joint European Association of Neuro-Oncology (EANO) and European Society for Medical Oncology (ESMO) guidelines for diagnosis, treatment and follow-up of patients with brain metastasis from solid tumors, any new neurological deficit in a cancer patient should always be suggestive of BM (4). Suspicious brain lesions may also appear on routine check-up magnetic resonance imaging (MRI)-scans of cancer patients, incidentally or during the recommended work-up (2). Singular lesions amenable to safe surgical resection should be operated upon, space-occupying lesions may even require urgent decompression (4, 5). Microsurgical tumor resection serves both therapeutic and diagnostic purposes, but at the risk of potential surgical complications particularly in frail patients (6).

Versatile histopathological and molecular-genetic analyses, however, should also be available in all unclear cases with multiple or highly eloquent lesions, particularly in patients with a history of more than one primary tumor, and those with unclear tumor status after therapy (7–9). Novel high-throughput sequencing methods have improved our understanding of individual cellular and molecular tumor targets. As a result, multiple novel personalized treatment strategies have been identified to treat cancer patients, thus opening novel treatment options for BMs. For example, in patients with Her2-positive breast cancer BMs (10, 11), those with ALK-rearranged (12, 13) or EGFR-mutated (14, 15) NSCLC BMs, and for BRAF V600 E mutated melanoma BMs (16), targeted therapies with significant intracranial activity are available. Still, there may be discrepancies between the actionable mutations of the primary tumor and their respective BM (17) and thus tissue-based analyses of BM can be necessary to guide therapy.

Due to the high recurrence rate of BM, follow-up imaging with short intervals is pivotal to monitor the course of disease and to potentially re-adjust therapy in case of tumor progression. However, suspicious lesions on MRI-scans can also be a manifestation of post-therapeutic changes, e.g., tissue necrosis after a radiation procedure or inflammatory reactions during immunotherapies, also termed pseudoprogression (18, 19). Due to similar visual characteristics, correct differentiation from tumor recurrence can be a diagnostic challenge. The response assessment in neuro-oncology (RANO) working group recommends O-(2-¹⁸Fluorethyl)-L-tyrosine ([¹⁸F] FET PET) to discriminate true tumor progression from pseudoprogression (20–22). Nevertheless, in unclear cases tissue acquisition remains the gold standard to resolve this diagnostic quandary and to select the appropriate treatment modality (18, 23).

Consequently, minimally invasive biopsy techniques are of high importance in the field of brain metastases (4, 5). Even though stereotactic frame-based biopsy represents a well-established procedure, general analyses of BM biopsy cases and their respective histopathologic results have only been performed in a few studies to date. Importantly, these studies have mostly lacked in-depth molecular data and concomitant analyses of the associated risk profile. With the present study, we aim to delineate diagnostic accuracy, intervention-related risks and the diagnostic benefit of stereotactic biopsy for suspected BM.

In this retrospective analysis from a high-volume comprehensive cancer center, we addressed the diagnostic value and peri-procedural risk of a highly standardized, advanced imaging-based stereotactic biopsy technique. Furthermore, we performed extended molecular-genetic analyses in a sub-cohort. Overall, the diagnostic accuracy of representative tissue samples was found to be high and the associated risk was low, even in highly eloquent locations such as the brain stem. The high diagnostic certainty of >98% definite neuropathological diagnoses (only 4 inconlusive cases among 234 biopsies for suspected metastases) and low peri-procedural risk of 2.6% for clinically relevant transient morbidity is in line with our previous results on the value of stereotactic biopsy in a large cohort of primary brain tumors (23), and differs from retrospective analyses by other groups studying the respective diagnostic yield (up to 11% inconclusive results) (29).

The low procedural risk and high diagnostic yield of the collected tumor tissue is realized due to the combination of two relevant factors. First, a spatially precise fusion of advanced high-resolution imaging data (including MR-angiography and PET) to the frame-based CT-scan. Second, a versatile, small-sample size optimized neuropathological evaluation integrating intraoperative smear-preparation for representative tissue selection. Because of the low bleeding rate, we have largely eliminated postoperative cranial CT scans from our clinical routine and limit it to the rare cases with diagnostic uncertainty to rule out a missed biopsy.

No comparison was made to frameless biopsy procedures. At our institution, the latter technique is usually applied only for superficial primarily dural lesions without significant involvement of adjacent brain tissue and for extended cortical-subcortical tissue cubes when vasculitis is suspected. There are no prospective studies addressing the different biopsy techniques in terms of diagnostic yield and associated risk profiles. However, retrospective studies have demonstrated that frameless biopsy also provides good diagnostic value with low procedural risk (30). Whether this is also the case for highly eloquently located lesions in the midbrain or brainstem has not been clearly shown. Indeed, eloquent location was associated with in increased risk of periprocedural morbidity in 284 cases undergoing frameless biopsy (31). In our clinical experience, this subgroup of patients is often referred to us for further evaluation from other university and/or tertiary centers. In our hands there is no obvious disadvantage in terms of time of operating theater occupancy and staff retention compared to frameless procedures (30): Our stereotaxy system (Brainlab^® Elements Stereotactic Planning) already enables target-point-accurate trajectory planning the day before, which is merely supplemented by the information from the intraoperative CT. The actual operating time is usually 20 minutes. A major advantage of frameless systems, however, lies in the prevention of intraoperative radiation exposure.

The intraoperative presence of the neuropathologist certainly contributed to the high quality of our result. Although the results of the smear preparations did not result in a second trajectory being performed, the neuropathologist can help to minimize the total number of serial biopsies needed by providing early feedback, thereby reducing the overall risk of the procedure (32). This could be of particular benefit in highly vascularized tumors and in the case of highly eloquent tumor localizations such as the brainstem.

The study population reflects the current challenges in patients with BM. In this large cohort of over 450 patients in 5 years, we demonstrate that approximately 50% were not amenable for surgical resection, but were referred for biopsy as part of a risk-adapted interdisciplinary treatment regimen. Of note, only BM patients referred to our neuro-oncology center due to diagnostic uncertainty were included in this study. In clinical routine, many BM patients with a limited number of small BMs in known primary tumors as well as those with miliary seeding are usually scheduled for radiosurgery, stereotactic fractionated protocols, or whole-brain irradiation without being discussed in an interdisciplinary tumor board. The majority of our study patients underwent stereotactic biopsy in the setting of newly diagnosed brain metastasis. In 40% of these patients, BM biopsy was recommended to diagnose the systemic tumor because systemic biopsy was deemed either technically impossible or too risky. Remarkably, in a small subset of patients with newly diagnosed suspected brain metastasis (5/114, 4.4%), a previously unknown second tumor was detected.

After BM treatment, routine follow-up imaging is recommended in short intervals to readily detect tumor progression and to re-adjust treatment recommendations accordingly. However, the differentiation of tumor relapse from pseudoprogression/radionecrosis still represents a major challenge in BM. Standardized MRI as well as [¹⁸F] FET PET is routinely performed at our institution according to current RANO guidelines (20, 21). However, the diagnostic certainty of [¹⁸F] FET PET outlined in this study (sensitivity 87.5%, specificity 36.4% to detect malignant progression) is not sufficient to guide therapy decisions, so that the indication for tissue diagnosis has to be confirmed. In fact, reactive alterations without significant tumor cell content were observed in a striking 50% of patients and pseudoprogression could be confirmed due to the subsequent clinical course of disease in all these cases. The rate of reactive alterations may further increase if treatment approaches combining radiotherapy and immunotherapy are applied. However, this combination was rarely administered in this series, and as a result no such analysis could be performed. In our neuropathological diagnosis, the transition from reactive changes in the sense of a pseudoprogress to (symptomatic) radiation necrosis appears to be fluid. In the absence of clear neuropathological differentiation criteria, the interpretation often depends additionally on the clinical appearance and the image morphological findings and remains an individual decision. High numbers of radionecrosis, however, were reported in a case series of 2,200 BM patients treated with radiosurgery (33). Follow-up investigation confirmed a recurrence in 203 cases (46%), radionecrosis in 118 cases (27%), both recurrence and radionecrosis in 30 cases (6.8%), and 90 patients (20%) displayed inconclusive results. An even higher number of 69% histologically confirmed cases of radionecrosis were reported in 35 BM after radiosurgery (34). Therefore, STX as a minimal-invasive tissue sampling procedure for accurate tissue diagnosis will certainly gain increasing relevance in the era of precision medicine for BM (35–37).

The evolving landscape of effective targeted therapies has significantly altered the management paradigm of BMs (7, 38, 39). For example, targeted therapies have established intracranial activity in patients with Her2-positive breast cancer BM (10, 11), ALK-rearranged (12, 13) or EGFR-mutated NSCLC BM (40) and for BRAF V600E mutated melanoma BM (41). For subgroups of asymptomatic patients, targeted systemic therapy as monotherapy even represents a first-line consideration (41–43). Notably, tumor-dependent discrepancies can arise between the actionable mutational profile of the primary tumor and the respective BM (17). Strikingly, approximately 50% of brain metastases can harbor clinically relevant mutations that are not present in the primary tumor, indicating significant clonal heterogeneity across the various geographic regions of the tumor (44). Therefore, tissue-based analyses of BM are not only important to understand the pathogenesis of tumorigenesis, but are essential in guiding therapeutic concepts. Discordance in regards to EGFR status between brain metastases and matched NSCLC samples has been reported in 0–33% of cases, whereas the discordance rate for ALK rearrangements lies in the range of 0–13% (8). For breast cancer BM, a discordance rate of 14% for Her2 and 29% for ER/PR has been reported (45). Discordant molecular profiles were also observed in a small subgroup of breast cancer patients in this case series. Such discrepancies indicate a dynamic, clonal evolution of the spreading disease and has important implications for combinatorial treatment approaches (46). In this context, a safe and simple way to diagnose and longitudinally evaluate BM is of increasing clinical relevance.

In summary, the high diagnostic yield and low complication rate supports an important role for minimal-invasive biopsy procedures in risk-adapted management algorithms for BM. Since it is still an invasive intervention, a reasonable and cautious assessment of the individual indication and risk-benefit profile is clearly demanded. However, due to increasingly specialized teams and interdisciplinary cooperation, a high-quality standard of this procedure can be maintained. While other diagnostic methods, such as liquid biopsy, represent a less invasive examination method, they are less-researched, still of experimental nature in most cases, and do not have the same informational value as stereotactic biopsy (47, 48).

Our study has several important limitations. Due to the retrospective study design, several relevant questions such as the significance of [18F] FET-PET, timing of biopsy, and longitudinal treatment data, remain unanswered and warrant future systematic study. Important information concerning the intraoperative interaction between the stereotactic neurosurgeon and treating neuropathologist regarding the number and use (for smear preparation vs. final neuropathologic assessment or molecular genetic analysis) of serial tissue samples cannot be objectively recorded. In addition, no qualitative comparison can be made with other biopsy techniques, such as frameless procedures. In our study, the result of neuropathologic examination was the gold standard and the basis for any management decision in individual cases. Although in all cases the further clinical course supported a correct assessment, clinical misjudgment based on neuropathologic diagnosis cannot be excluded with absolute certainty.

In conclusion, image-guided stereotactic biopsy represents a valid and safe tool for diagnosis and even molecular characterization of BM. The precise identification of the molecular signatures of BM can guide the appropriate choice of targeted therapies, heralding a new era of precision medicine in the treatment of primary and recurrent brain metastases.

The datasets presented in this article are not readily available because of national and institutional laws to protect patient confidentiality. Requests to access the datasets should be directed to the Center for Neuropathology and Prion Research of the University Hospital of Munich.

Ethical approval for this analysis was obtained from the ethics committee of the Ludwigs-Maximilians University Hospital (project number 22-0476). Patients provided informed written consent to allow for anonymous or pseudonymous data handling.

J-CT, SQ, LB and NT contributed to conception and design of the study. AD, SK, SQ, and NT organized the database, evaluated the clinical courses and performed image analyses. AD and SK carried out the statistical analyses. AD, SQ, SK, JW, J-CT, LB, and NT wrote the manuscript. MD, NA, RF, MN, and RE edited the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.