This paper was preceded by our previous publication by Yuen et al in 2014, with an emphasis on the diagnostic accuracy and an aim for improvement.[26] We previously reported a diagnostic accuracy of 93.0% (50/54) and symptomatic haemorrhage of 1.9% (1/54) in a series of 54 frameless stereotactic brain biopsy procedures. The current study was a continuation of the previous study with an accumulation of 109 procedures with same criteria for diagnostic accuracy. In our current study, the diagnostic accuracy was 94.5% (103/109) with a symptomatic hemorrhage rate of 0.9% (1/109). The diagnostic accuracy of frameless stereotactic brain biopsy among the relatively large case series with procedure numbers more than 100 in the recent 10 years were shown in Table 7. The diagnostic yield of 94.5% in this study was within the range of 72.8–99.7% and the accuracy was pronounced considering the structurally defined criteria for diagnostic accuracy in the current study. [23, 8, 2, 22, 18, 13, 24, 17, 14, 1, 10, 7]
Table 7
Comparison of Frameless brain biopsy results from different neurosurgical centers with frame-less stereotactic biopsy performed
Author | Center | Year | Number of biopsy/patients | Navigation | Diagnostic Accuracy | Symptomatic hemorrhage | Mortality |
Current study | Hong Kong | 2021 | 109 | BrainLab, Fiagon | 94.5% | 0.9% | 0.0% |
Ungar et al [23] | Israel | 2021 | 130 | Medtronic StealthStation | 96.9% | 1.6% | 0.8% |
Giamouriadis et al [8] | United Kingdom | 2019 | 371 | Medtronic AXIEM | 94.1% | 0.9% | 1.1% |
Æbeløa et al [2] | Denmark | 2019 | 111 | Medtronic Stealthstation | 98.2% | 3.6% | 1.8% |
Sciortino et al [22] | Italy | 2019 | 140 | BrainLab | 93.6% | 2.1% | 0.0% |
Mader et al [18] | Germany | 2018 | 119 | BrainLab | 92.4% | 1.7% | 0.8% |
Kellermann et al [13] | Germany | 2017 | 229 | Stryker STP3 | 91.7% | 0.4% | 0.4% |
Verploegh et al [24] | Netherlands | 2015 | 247 | Medtronic StealthStation, BrainLab | 94.6% | 1.2% | 0.8% |
Lu et al [17] | USA | 2015 | 113 | GE Healthcare | 89.4% | N/A | N/A |
Khatab et al [14] | Netherlands | 2014 | 235 | Medtronic StealthStation, Zeiss-MKM, | 72.8% | 2.1% | 0.9% |
Air et al [1] | USA | 2012 | 284 | Medtronic StealthStation, BrainLab | 89.2% | N/A | N/A |
Harrisson et al [10] | United Kingdom | 2012 | 150 | Medtronic StealthStation AxiEM | 96.7% | 4.7% | 2.7% |
Frati et al [7] | Italy | 2011 | 296 | Medtronic StealthStation | 99.7% | 1.0% | N/A |
Notes: N/A – Not available from the published journal. |
The goal of the current study was to seek improvement in the future biopsy procedures. In our case series, there were four inconclusive pathological results and two non-diagnostic pathological results. With respect to Yuen et al in 2014, while no more symptomatic haemorrhage case occurred, there were one inconclusive and one negative results encountered after 2014.[26] We further analyzed all the six procedures in detail with reference to the target selection, trajectory planning, intra-operative navigation, aspiration technique, and post-operative biopsy tract verification. The identified causes for inconclusive or negative diagnosis included two inappropriate target selection, three inaccurate targeting and one incorrect timing, as shown in Table 5. All these errors have been described previously and the question to the neurosurgeons is how to avoid these errors.[16, 19, 9]
One of the inconclusive pathology results in 2007 was due to incorrect time to biopsy (Table 5, Case 1). It was a case of primary central nervous system lymphoma (PCNSLP) who received steroid prior to the biopsy (Fig. 1). It was well known that PCNSLP was very sensitive to steroid. For patients with brain lesion undergoing biopsy with steroid therapy commenced, an updated imaging is very important. If there is significant shrinkage of the lesion, one important differential diagnosis is lymphoma and in such case, the biopsy should be withheld.[4] We learned the lesson and no similar error occurred in the subsequent 30 cases of suspected primary CNS lymphoma.
There were two inconclusive pathologies occurred in 2016 and 2017 respectively due to error in target selection (Table 5, Case 4 and 6). In one procedure, the target was the central necrotic area and the other was inside the tumour cyst. Retrospectively, the error might have been avoided if we paid more attention to the characteristic of the imaging finding (Fig. 2 and Fig. 3).
There were three nondiagnostic procedures with inaccurate targeting evidenced on the post-operative CT. All lesions were in the parietal lobe near the trigone area. In one of the cases the error could also be contributed to its small size and the proximity to the trigone of the lateral ventricle (Fig. 4). Birski et al reported nondiagnostic yield of 5/85 intraventricular or periventricular lesions.[3] Our experience was that using facial recognition for registration, special attention should be paid to the accuracy in the parieto-occipital region. The accuracy can be checked by the function of the navigation software and improved by adding registration points to the parieto-occipital scalp during the registration procedure. Firmly fixed skull pins and the refractory array are also important points in order to avoid the inaccuracy in targeting.
Reviewing the six cases without a diagnostic pathology result, we believed that the error could be definitively avoided in the timing of biopsy for lymphoma and could likely to be avoided in cases with inaccurate targeting. For the two cases with target selection error, avoidance remained possible.
Apart from a positive histological diagnosis for tumour type and grading, further immunohistochemical and molecular evaluations were increasingly important in the diagnosis and prognosis of certain types of brain tumors. For patients with GBM, the absence of TERT mutation and the presence MGMT promoter methylation had better prognosis.[6, 11] Mader et al reported the nondiagnostic rate of 1.5% (1 in 67 cases) for MGMT promoter methylation among GBM cases in 2018.[18] We have three GBM cases where immunohistochemical staining was incomplete due to insufficient tumor sampling, including two GBM cases with TERT mutation detection not performed and one GBM case while test for IDH1 mutation was negative, other immunohistochemical test was incomplete. Weise et al suggested that the tangential trajectory planning along the long axis of the lesion would enhance the MGMT promoter methylation detection.[25] Our experience was that a minimum of four aspirations with appropriate size tissues together with a good communication with the pathologist intra-operatively for the quality and quantity of the biopsied tissue, were necessary for an adequate pathological and immunohistochemical diagnosis.
Another important topic for stereotactic brain biopsy was to examine the agreement of pathological diagnosis between the biopsy and open surgery. Discrepancy between stereotactic biopsy and open surgery had been reported to be 38% in Jackson et al, 24% in Reithmeier et al and relatively low (3.3%) in Kim et al.[12, 21, 15] The reason for discrepancy between stereotactic biopsy and open surgery was that the biopsied site might not be fully representative of the entire lesion and the true diagnosis might not be reached, which would affect the subsequent disease course and treatment.[12] Among the 11 cases that received both biopsy and open surgery in our case series, there were two cases with discrepancy in the biopsy and subsequent excision result and thus the reliability rate was 82%. These two cases were known to be non-diagnostic after the stereotactic biopsy. We believed that appropriate target selection and post-operative auditing could improve the reliability and avoid this discrepancy.
The symptomatic hemorrhage rate in our study was 0.9% due to one hemorrhage that occurred in 2009 when we used 2.5mm diameter biopsy needle. Our current study enhanced the believe that the rate of symptomatic haemorrhage was low without compromising the diagnostic accuracy with the 1.8mm diameter biopsy needle.
There were certain limitations in our research. Firstly, this was not a prospective study though stereotactic brain biopsy procedure was collected based on the operating protocol at the time of surgery for each procedure. Considering a study of 14 years interval, experience accumulation over time might weaken the comparability between early and recent results. Nevertheless, our analysis for the diagnostic accuracy and illustration on the nondiagnostic cases provided insights for further improvements for frameless stereotactic brain biopsy.