Efficacy of a Second Brain Biopsy for Intracranial Lesions after Initial Negativity

Patients

We retrospectively reviewed the medical records and histology reports of all adults who underwent a brain biopsy at La Pitié-Salpêtrière University Hospital between January 1, 2009 and December 31, 2019. Patients who underwent two distinct brain biopsies for histological analyses of the same lesion—where the first biopsy was nondiagnostic—were included in the analysis. Patients with incomplete data were excluded.

Study variables and outcomes

The main outcome variables were 1) obtaining a specific histological diagnosis and 2) brain-biopsy-related complications. The other analyzed variables included demographic characteristics, medical history, clinical manifestations, laboratory findings, and brain MRI findings before performing the initial and second brain biopsies.

The histological results of the brain biopsies were categorized into three groups: specific histological diagnosis, nonspecific histology, and normal brain. Obtaining a specific histological diagnosis was defined as the brain-biopsy findings of a specific lesion being sufficient by themselves to make a diagnosis and to modify the therapeutic management.

A multidisciplinary discussion involving neurosurgeons, pathologists, neuroradiologists, neurologists, and internists determined whether a brain biopsy of a nonspecific lesion(s) contributed to the final diagnosis or whether a second biopsy was required to obtain a specific histological diagnosis. The participants in these discussions systematically and comprehensively reviewed the following aspects for each patient: medical history, neurological and extraneurological findings, less-invasive diagnostic workup, and initial brain-biopsy histological results. Two senior neuroradiologists analyzed all of the imaging findings, including available MRI data [T1-weighted fluid-attenuated inversion-recovery (FLAIR) sequences, T2-weighted FLAIR sequences, T1-weighted sequences with gadolinium injection, gradient-echo T2^*-weighted sequences, and diffusion-weighted sequences] and multiparametric imaging data. Two senior neuropathologists examined all of the histological slides. A second biopsy was proposed if no consensus was reached during the multidisciplinary discussion about whether the results obtained from the first brain biopsy contributed to a diagnosis.

We attempted to classify the reason why the first biopsy was nondiagnostic into three categories: target error, interpretation error, or sample-size error. The location of the biopsy target was assessed by merging the images from the postbiopsy CT scan with those from prebiopsy MRI.

Brain-biopsy-related complications were defined as occurring during the month following the procedure. We recorded the symptomatic complications as well as the asymptomatic hemorrhages that were visible only on a postoperative CT scan.4, 5

Surgical procedures and postoperative management

The patients who required corticosteroids before the biopsy initially received methyl prednisone at 2 mg/kg/day for 2 days, and then the daily dose was reduced by half every 2 days until the day on which the biopsy was performed. A stereotactic biopsy was used for deep-seated lesions according to a previously described procedure.6 Six to eight biopsy samples (~10 mm) were collected for both the primary and secondary stereotactic procedures. Biopsy samples for cortical lesions were obtained via an open craniotomy or a burr hole. We considered a gold-standard diagnostic open-biopsy sample to comprise 1 cm³ of tumoral cortex.

The collected tissue samples were sent for histopathological analyses. When the differential diagnosis included infection, tissue was set aside for microbiology investigations. Patients were monitored for at least 6 h in the recovery unit. Prior to transfer to the neurosurgery department, a postoperative CT scan was systematically obtained to rule out immediate complications.

Statistical analyses

Results expressed as number and percentage values were compared using χ² tests, while continuous variables expressed as mean±SD or median (interquartile range) values were compared using Student's t-test. The demographic, clinical, radiological, and biological characteristics of the patients were tested in univariate analyses for their associations with obtaining a specific histological diagnosis, based on intraindividual comparisons between the first and second biopsies. Thereafter, multiple logistic-regression analyses using backward stepwise variable elimination were performed, with the variable-exit threshold set at p>0.10. Factors for which p≤0.10 in our univariate analyses were entered into the multivariate model. Statistical significance was defined as p<0.05. Analyses were performed using IBM SPSS Statistics software (version 21, IBM Corp., Armonk, NY, USA).

Standard protocol approvals, registrations, and patient consents

The database utilized in this study is registered with the Commission Nationale de l'Informatique et des Libertés (no. 2214386). In accordance with the ethical standards of the Institutional Review Board of our hospital (8610Z-MR3), the Committee for the Protection of Human Subjects, and French law, written informed consent was not needed for the current demographic, physiological, and hospital-outcome data analyses because this observational study did not modify existing diagnostic or therapeutic strategies; however, all of the patients were informed about their inclusion in the study. The manuscript was prepared in accordance with the STrengthening the Reporting of OBservational studies in Epidemiology statement.

Data availability statement

Anonymized data will be shared upon request from any qualified investigator.

Study population

During the study period, 1,919 patients underwent a brain biopsy, and the 30 who were biopsied twice (1.6%) were included in the analysis (Fig. 1). The male-to-female ratio was 0.76. The age on the first biopsy day was 53.9±16.1 years (range=22.5–73.6 years), the median time interval between the two biopsies was 63 days (range=12–1,064 days, and the follow-up after the first biopsy had a duration of 32±41 months (range=2–180 months).

Fig. 1
Flow chart of patient inclusion in this study of the contribution to the diagnosis of a second brain biopsy after a nondiagnostic first biopsy.

• Click for larger image
• Download as PowerPoint slide

The general characteristics of the 30 analyzed patients and their brain biopsies are presented in Table 1. The initial clinical manifestations included neurological deficit (50.0%), extraneurological symptoms (10.3%), seizures (13.8%), and fever (10.3%). Most patients had unifocal (73.3%) lesions. The biopsy-targeted initial lesions were predominantly supratentorial (93.3%), deep-brain-seated (86.7%) with a largest diameter >1 cm (96.7%), and gadolinium-enhanced (73.3%) lesions. All of the initial biopsies were performed stereotactically, with MRI guidance performed in 80.0% of them.

Table 1
Patient and biopsy characteristics

• Click for larger image
• Click for full table
• Download as Excel file

Diagnoses

All of the 30 patients analyzed in this study had initially undergone a nondiagnostic brain biopsy, and the second biopsy obtained a specific histological diagnosis for 26 (86.7%) of them [odds ratio (OR)=7.5, 95% confidence interval (CI)=3.0–18.7, p<0.001]. These 26 patients comprised 16 (61.5%) with glioma, 4 (15.4%) with lymphoma, 4 (15.4%) with metastasis, 1 (3.8%) with nocardiosis, and 1 (3.8%) with radionecrosis (Table 2). The negative result of the first biopsy was attributed to a target error in ten (38.5%) patients, a sample-size error in ten patients (38.5%), and an interpretation error in six (23.0%) patients.

Table 2
Histological diagnoses obtained from the first and second brain biopsies

• Click for larger image
• Click for full table
• Download as Excel file

The detailed clinical, radiological, and histological features of the patients are presented in Table 3. Fig. 2 illustrates the differences between the nondiagnostic first biopsy and the diagnostic second biopsy in four representative cases.

Fig. 2
Four example cases illustrating the differences between the nondiagnostic first biopsy and the diagnostic second biopsy. The blue line indicates the biopsy target obtained by merging postoperative CT scan and preoperative MRI. Case #15. A 63-year-old female underwent a stereotactic brain biopsy for a left middle cerebellar peduncle lesion. The initial neuropathological examination found only normal cerebellar tissue. A postoperative CT scan showed a target error (A1). The neuropathological examination of the second stereotactic biopsy performed 19 days later was consistent with breast cancer metastasis (A2). The red arrowhead shows the target of the first biopsy (A2). Case #18. A 68-year-old male presented with left hemiparesis. Brain MRI showed a right frontal “flaky” contrast-enhanced lesion. Corticosteroids were introduced and then stopped 3 days before performing the stereotactic biopsy. The neuropathological analysis revealed inflammatory lesions with macrophage infiltration (B1). This noncontributory result led to a second biopsy being performed 25 days after withdrawing the corticosteroids, which led to a diagnosis of a B-cell primary lymphoma in the central nervous system (B2). Case #25. A 63-year-old male was admitted due to suspected left medial temporal tumor. The neuropathological examination of the stereotactic biopsy sample revealed nonspecific necrosis (C1). Corticosteroids were introduced after the biopsy, and tumor regression was found in brain MRI performed 20 days later (C2). Corticosteroid therapy was therefore stopped, and follow-up imaging revealed increases in the size and spread of the lesion (C3). A second brain biopsy performed 45 days after the first biopsy disclosed a grade IV glioma. Case #30. A 57-year-old female underwent a stereotactic brain biopsy due to suspected left medial temporal tumor (D1). The neuropathological examination revealed nonspecific necrosis. The second stereotactic biopsy performed 3 weeks later was consistent with a grade IV glioma (D2). The red arrowhead shows the particularly interesting finding of the trajectory of the first needle biopsy (D2).

• Click for larger image
• Download as PowerPoint slide

Table 3
Clinical, radiological, and histological features of the included patients

• Click for larger image
• Click for full table
• Download as Excel file

Factors associated with the diagnostic yield

Intraindividual comparisons of the demographic, clinical, radiological, and biological characteristics between the initial nondiagnostic biopsy and the second biopsy that resulted in a histological diagnosis are reported in Table 1. More than onethird (36.7%) of the patients had been taking corticosteroids before the initial biopsy, while 3.8% had been taking them before the second biopsy (p=0.003). The largest lesion diameter was significantly greater at the second biopsy (35.6 mm vs. 31.6 mm, p=0.005). The logistic-regression model used in the multivariate analyses showed that biopsy samples obtained from patients treated with corticosteroids during the prebiopsy period were more often nondiagnostic (OR=2.6, 95% CI=1.1–6.0, p=0.01). The diseases associated with nondiagnostic biopsies affected by corticosteroids were lymphoma (66.7%), glioma (16.7%), and nocardiosis (16.7%). Three (75.0%) of the four patients who had undergone a nondiagnostic second biopsy had still been taking corticosteroids before the second biopsy (Table 3). Apart from corticosteroids, there were no major changes in the medications taken by patients between the first and second biopsies that could have influenced the biopsy results.

Postbiopsy complications

None of the patients developed symptomatic complications after the first biopsy, while two (6.0%) patients experienced transient complications after the second biopsy (p=0.49) (Table 4). One patient experienced a generalized epileptic seizure a few minutes after the intervention, and another patient presented with postbiopsy hemiparesis that resolved fully after corticosteroid treatment. Asymptomatic hemorrhages related to biopsy procedures that were diagnosed only in systematic postbiopsy imaging were observed in 6 (20.0%) patients after the first biopsy and in 13 (43.3%) patients after the second biopsy (p=0.10).

Table 4
Postbiopsy complications

• Click for larger image
• Click for full table
• Download as Excel file

Key results

We found that performing a second brain biopsy in patients who had undergone an initial nondiagnostic biopsy was effective in 87% of cases. Prebiopsy corticosteroid treatment appears to be a predictor of a nondiagnostic biopsy.

Predictors of nondiagnostic biopsies

Needle biopsies can lead to diagnostic errors due to the presence of existing hemorrhagic products or necrosis, or when the biopsy fails to target the contrast-enhancing lesion.7 Therefore, performing an open biopsy rather than a needle biopsy should be the first step when it is possible, in order to avoid a nondiagnostic biopsy. For lesions of concern with a necrotic center, the biopsy should target the area of fine parietal peripheral contrast enhancement. For such cases we recommend performing multiple biopsies at different depths along the same trajectory in order to maximize the probability of success.

Many factors can lead to a nondiagnostic result. Lu et al.8 considered that prior lesion treatment (surgery or radiation) is strongly negatively associated with the biopsy diagnostic yield. However, because none of our patients had received treatment prior to the biopsy, we cannot comment on this previous finding.

Precision is the ultimate aim of a stereotactic biopsy, and its usefulness is dependent on minimizing errors at every step of the procedure, including frame application, image acquisition, image manipulation, surgical planning of the target and trajectory, patient positioning, and the surgical procedure itself. The neurosurgeon should be familiar with the principles of stereotactic techniques. Errors in target calculation can also occur due to image distortion;9 however, near-complete distortion correction can be reliably achieved with modern devices.4 Moreover, several studies have underlined the importance of the experience of the neurosurgeon in obtaining a histological diagnosis and reducing postbiopsy complications.1, 2, 10, 11 Lastly, the biopsy method (stereotactic or open) applied in our study was chosen according to the location of the lesion, and it did not influence the histological results.

Prebiopsy corticosteroid therapy is known to increase apoptosis,12, 13, 14, 15, 16 and it is considered responsible for misdiagnosing primary lymphomas of the central nervous system.17, 18, 19 This concept has recently been challenged by study results suggesting that the diagnostic yield for lymphoma patients receiving corticosteroids is comparable to those for biopsies performed without prebiopsy corticosteroid therapy.20, 21 Moreover, data obtained from in vitro experiments suggest that mineralocorticoids exert a variable, time-dependent inhibitory effect on the proliferation of glioma cells.22 Indeed, some authors have reported the regression of glial tumors in patients receiving corticosteroids.23, 24, 25, 26, 27, 28 However, the molecular mechanisms underlying the effects of corticosteroids on tumor cell regulation are still poorly understood.

In our series, prebiopsy corticosteroid therapy was the only risk factor for a nondiagnostic biopsy. In the light of our results and experience, we recommend performing brain biopsies in patients without corticosteroid therapy, even if this means planning them urgently in cases requiring corticosteroid administration without delay. In cases of suspected lymphoma, we consider that a 1-week corticosteroid-free therapeutic window is sufficient to avoid the risk of performing a nondiagnostic biopsy. However, this recommendation depends on the clinical condition of the patient, including since in some cases it can be difficult to wait 1 week without administering corticosteroids. Tailored management should therefore be developed that considers the trade-off between the risk of performing a nondiagnostic biopsy and the risk of clinical worsening by withdrawing corticosteroids.

Methods for enhancing biopsy efficacy

Various methods have been reported for reducing the rate of nondiagnostic biopsies. Mathon et al.6 applied intraoperative smears, which decreased the rate of nondiagnostic biopsies from 2.6% to 0%. Preoperative 5-ALA administration and the intraoperative assessment of fluorescence could improve the diagnostic yield of needle biopsies, especially for glioma and lymphoma patients.29, 30 Akshulakov et al.31 reported innovative intraoperative techniques for identifying viable pathological tissue, which included optical and molecular detection methods.

Avoiding a second biopsy by performing molecular testing of the first nondiagnostic sample

Intratumoral necrosis, hemorrhage, and peripheral gliosis are considered the main causes of nondiagnosing biopsies, and recent studies have shown that IDH1 and pTERT mutations can be detected in a high proportion of nondiagnostic biopsies from IDH1 and pTERT-mutant glioma patients,7, 32 which can help to avoid the need to perform an additional biopsy. The efficacy of this type of molecular analysis has already been demonstrated for biopsies of indeterminate thyroid nodules33 and when interpreting bronchoscopy findings in patients with suspected lung cancer.34 Moreover, genomic next-generation sequencing of brain tissue performed over the last few years has enabled diagnoses that were not possible using routine histological and microbiological testing.35 These recently reported novel techniques pave the way to further improve the diagnostic yield of brain biopsies.36

Limitations

This study had both limitations and strengths. First, it had a retrospective, monocentric, observational design, but intraindividual comparisons made it possible to eliminate most of the bias. Second, the patient population was quite heterogeneous regarding type of brain lesions, but this reflects the real-world situation of biopsied patients in a tertiary referral center.

Conclusion

Performing a second brain biopsy in patients who have undergone an initial nondiagnostic biopsy is effective in most cases. Given the significant difficulties in managing patients without a definitive diagnosis, we advocate that a second biopsy be systematically considered in the diagnosis algorithm of these patients after it has been verified that molecular testing cannot help to obtain a diagnosis. Corticosteroid administration can lead to nondiagnostic biopsies and should be avoided when possible during the prebiopsy period.