OUP user menu

Brain biopsy in dementia

J. D. Warren, J. M. Schott, N. C. Fox, M. Thom, T. Revesz, J. L. Holton, F. Scaravilli, D. G. T. Thomas, G. T. Plant, P. Rudge, M. N. Rossor
DOI: http://dx.doi.org/10.1093/brain/awh543 2016-2025 First published online: 18 May 2005

Summary

Brain biopsy has an uncertain role in the diagnosis of dementia. Here we report a retrospective analysis of 90 consecutive cerebral biopsies undertaken for the investigation of dementia in adults at a tertiary referral centre between 1989 and 2003. In most cases (90%), biopsy consisted of a right frontal full thickness resection of cortex, white matter and overlying leptomeninges. Fifty-seven per cent of biopsies were diagnostic: the most frequent diagnoses were Alzheimer's disease (18%), Creutzfeldt–Jakob disease (12%) and inflammatory disorders (9%). Other diagnoses in individual patients included Pick's disease, corticobasal degeneration and other tauopathies, Lewy body dementia, multiple sclerosis, Whipple's disease, progressive multifocal leucoencephalopathy, cerebral autosomal dominant arteriopathy with subcortical ischaemic leucoencephalopathy, vasculopathies and paraneoplastic encephalopathy. The most frequent biopsy finding in the non-diagnostic group and for the series as a whole (37%) was non-specific gliosis variably affecting both cortex and white matter. Complications (11%) included seizures, intracranial and wound infections, and intracranial haemorrhage; there were no deaths or lasting neurological sequelae attributable to the procedure. No trends in diagnostic yield or complication rate over the course of the series were identified. Information obtained at biopsy determined treatment in 11%. A raised cerebrospinal fluid cell count was the only robust predictor of a potentially treatable (inflammatory) process at biopsy. The constellation of behavioural change, raised CSF protein and matched oligoclonal bands in CSF and serum was associated with non-specific gliosis at biopsy. This series underlines the value of cerebral biopsy in the diagnosis of dementia, and suggests that certain clinical and laboratory features may be useful in guiding the decision to proceed to brain biopsy where a treatable disease cannot be excluded by other means.

  • brain biopsy
  • dementia
  • Alzheimer's disease
  • non-specific gliosis
  • CADASIL = cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy
  • CJD = Creutzfeldt–Jakob disease

Introduction

Accurate diagnosis of dementia is frequently difficult, and definitive diagnosis during life may depend on the examination of tissue obtained at brain biopsy. This is an urgent clinical problem in cases where a treatable cause cannot be excluded by non-invasive means. However, brain biopsy is generally regarded by clinicians as a procedure of last resort, if it is used at all: this is due both to the perceived low diagnostic yield of the procedure and the possibility of serious complications, including anaesthetic catastrophes, haemorrhage, infection, seizures and death (Biemond, 1966; Blackwood, 1970; Hulette et al., 1992; Scolding et al., 1997; Hauw and Duyckaerts, 2002). The procedure has been variously described as a necessary evil (Biemond, 1966), condemned as unethical (Hauw and Duyckaerts, 2002) and recommended as a valuable aid to diagnosis (Blackwood and Cumings, 1959; Wagner and Wisotzkey, 1963). The decision to undertake brain biopsy depends on a clinical risk : benefit analysis, in which the probability of obtaining a diagnosis and possibly altering management must be weighed against the risks of the procedure and the hazards of empirical therapy.

Brain biopsy is typically considered in younger patients where clinical or laboratory features raise the possibility of a reversible (usually inflammatory) process, in particular isolated vasculitis of the CNS. This disorder may present as insidious cognitive decline, and diagnosis remains notoriously difficult (Parisi and Moore, 1994; Chu et al., 1998). No consensus exists regarding the value of brain biopsy in suspected vasculitis, although it is likely to be superior to conventional catheter angiography (Vollmer et al., 1993; Chu et al., 1998; Alrawi et al., 1999), which is associated with significant false-positive and false-negative rates. In addition to primary vasculitis of the CNS, a range of other potentially treatable processes, including systemic vasculitides (Parisi and Moore, 1994; Chu et al., 1998), infections such as Whipple's disease (Ryser et al., 1984; Brown et al., 1990), inflammatory conditions such as Sjögren's syndrome (Caselli et al., 1991) and neurosarcoidosis (Hayashi et al., 1995), neoplasms (LeWitt et al., 1983; Arboix et al., 2000) and vasculopathies (Larner et al., 1999) may present predominantly or exclusively with dementia and may elude diagnosis by non-invasive means.

A number of published studies have evaluated brain biopsy in the investigation of adult patients with dementia (Green et al., 1952; Blackwood and Cumings, 1959, 1966; Wagner and Wisotzkey, 1963; Eadie, 1964; Sim et al., 1966; Smith et al., 1966; Groves and Moller, 1966; Moossy, 1969; Coblentz et al., 1973; Torack, 1979; Kaufman and Catalano, 1979; Bowen et al., 1982; Neary et al., 1986; Katzman et al., 1988; Waltregny et al., 1989, 1990; Hulette et al., 1992). The procedure has also been assessed in the context of concomitant shunt surgery for suspected normal pressure hydrocephalus (Savolainen et al., 1999; Golomb et al., 2000) or prior to experimental ventricular infusion therapy (Harbaugh et al., 1989; DeKosky et al., 1992). Series for which adequate information is available are summarized in Table 1. The usual site of brain biopsy for dementia is the non-dominant (right) frontal lobe; no consensus has emerged concerning the value of biopsy targeted to focal brain lesions. Published series vary widely in clinical composition, indications for biopsy, surgical and pathological techniques and duration of follow-up. In the studies summarized in Table 1 a specific diagnosis was achieved in 22–84% of brain biopsies; however this variation in accuracy includes clinically typical cases where biopsy was performed to assess the accuracy of clinical diagnosis rather than to resolve diagnostic uncertainty (for example, in the series of DeKosky et al., 1992, patients were selected according to consensus guidelines for Alzheimer's disease). The most common pathological diagnosis across studies is Alzheimer's disease. A number of other pathologies are represented, including Creutzfeldt–Jakob disease (CJD), degenerative dementias (including Pick's disease, dementia with Lewy bodies and congophilic angiopathy), inflammatory disorders (including cerebral vasculitis, chronic meningoencephalitis and granulomatous disease), cerebrovascular disease, multiple sclerosis, neoplasia, leucodystrophies, storage diseases, undetermined encephalopathies and non-specific gliosis. Up to 72% of biopsies showed non-diagnostic abnormalities, and up to 55% of biopsies were described as showing no abnormalities. The incidence of complications in these studies ranged from 0 to 14%. Serious complications were rare, and included intracranial haemorrhage (Green et al., 1952; Smith et al., 1966; Hulette et al., 1992), subdural hygroma (Hulette et al., 1992), transient hemiparesis (Neary et al., 1986), focal seizures (Kaufman and Catalano, 1979), wound infection (Green et al., 1952; Eadie, 1964), prolonged confusional state (Neary et al., 1986), aspiration pneumonia and pulmonary embolism (Kaufman and Catalano, 1979). The overall mortality associated with diagnostic brain biopsy in dementia has been estimated as approximately 1% (Katzman et al., 1988; Chen et al., 1996).

View this table:
Table 1

Summary of studies of brain biopsy in dementia

AuthorNo.Age range (years)Diagnostic abnormalitiesNon-diagnostic abnormalitiesNo abnormalitiesMost frequent diagnosisOther findingsComplications (%)
No.%No.%No.%No.%Nature
Green et al., 195215NS853747ADPiD, undetermined encephalopathy211Haemorrhage, wound infection
Blackwood and Cumings, 19599NS778222ADVascular disease, MS
Eadie, 19649>40333444222Vascular diseaseAD, non-specific abnormalitiesNS14*Wound infection
Smith et al., 19665930–6942711729ADPiD, vascular disease, spongiform encephalopathy, chronic meningo-encephalitis, non-specific12Subdural haematoma (one death)
Coblentz et al., 19731547–63128021317ADPiD, microinfarcts, unclassifiableUncertain
Torack, 197929NS16551345ADCJD, encephalitis, congophilic angiopathyNS
Kaufman and Catalano, 19791828–69422137215CJDChronic meningo-encephalitis, MS, astrocytosis, perivasculitis, haemorrhagic encephalopathy, neuronal inclusions, granulomataNS13*Pneumonia (one death), pulmonary embolism, seizures
Bowen et al., 19823635–69267226822ADVascular disease, CJD, leucodystrophy, microgliomaNS
Neary et al., 19862443–691875625ADNon-specific pathology28**Transient hemiparesis, persisting confusion
Katzman et al., 198823NS1983417ADCJD, microinfarcts, PiD, chronic meningitis, isolated CNS vasculitis313Subdural haematoma, aseptic meningitis
Waltregny et al., 1989, 199028†† (246)23–69227941427ADCJD, non-specific gliosis, Kufs disease, metachromatic leucodystrophy27Wound infection
Hulette et al., 19921432–78857429214CJDAD, DLB, Niemann–Pick type 2, anaplastic astrocytoma, non-specific gliosis214Intracranial haemorrhage, subdural hygroma
Harbaugh et al., 19894543–853884716ADPiD, congophilic angiopathyUncertain
DeKosky et al., 1992§
Savolainen et al., 1999118NS4740656555ADNon-specific reactive gliosisUncertain
Golomb et al., 200056NS234112212138ADDiffuse plaquesNS
Present study9016–745157384211ADSee Table 31011Seizures, intracerebral haemorrhage, wound infection, cerebral abscess, subdural haematoma, pneumonia
  • AD = Alzheimer's disease; CJD = Creutzfeldt–Jakob disease; DLB = dementia with Lewy bodies; MS = multiple sclerosis; NS = not stated; PiD = Pick's disease.

  • * Estimated from larger biopsy series;

  • retrospective study;

  • concurrent ventricular shunting for possible normal pressure hydrocephalus;

  • § concurrent ventricular catheter placement;

  • ** excluding transient confusion or pyrexia;

  • †† multiple stereotactic biopsies in each patient.

Most authors have stressed the value of brain biopsy for accurate diagnosis in dementia, due to the difficulty of distinguishing reversible from irreversible processes on clinical or other grounds (Green et al., 1952; Blackwood and Cumings, 1959; Eadie, 1964; Sim et al., 1966; Torack, 1979; Kaufman et al., 1985; Neary et al., 1986; Katzman et al., 1988; Hulette et al., 1992; Salerni and Schmidek, 1995; Chen et al., 1996). Other suggested benefits include the provision of information to the patient's relatives and the avoidance of additional procedures (Eadie, 1964; Biemond, 1966; Groves and Moller, 1966; Blackwood, 1970; Kaufman and Catalano, 1979; Hulette et al., 1992). Contentious issues include the safety of the procedure, its practical value in guiding treatment and appropriate timing in relation to clinical course (Blackwood and Cumings, 1959; Biemond, 1966; Groves and Moller, 1966; Torack, 1979; Kaufman and Catalano, 1979; Paulson, 1983; Neary et al., 1986; Katzman et al., 1988). Authors have discussed a number of factors that limit the diagnostic utility of brain biopsy, including artefactual change associated with the procedure, the range of non-specific histological abnormalities of uncertain pathological significance, and focal or patchy tissue pathology that may not be captured at biopsy (Smith et al., 1966; Kaufman and Catalano, 1979). Other caveats may include the difficulty of obtaining the patient's informed consent, limited experience in the procedure itself and in the analysis and interpretation of pathological findings at any single institution, and the risk of transmission of disease (Kaufman and Catalano, 1979; Hulette et al., 1992).

The period spanned by these earlier studies has seen considerable evolution in diagnostic modalities and surgical and histopathological (including immunohistochemical) techniques, and the appearance or recognition of new diagnostic entities. The relative impact of recent techniques on the diagnostic value of brain biopsy for dementia remains unclear, yet information about potential risks and benefits is necessary if patients are to give informed consent to undergo the procedure. The role of brain biopsy in dementia is anticipated to become an increasingly important issue with the advent of disease-modifying therapies for neurodegeneration. However, the frequency of treatable causes of dementia appears to be declining in recent series (Piccini et al., 1998), probably reflecting earlier and more accurate clinical diagnosis, more sensitive brain imaging techniques and the widespread use of routine screening investigations (Marsden and Harrison, 1972; Kaufman and Catalano, 1979). It is therefore timely to re-evaluate the role of brain biopsy in the investigation of dementia. In order to address this issue, we performed a retrospective analysis of all cases of cerebral biopsy for dementia undertaken at a large tertiary referral centre (National Hospital for Neurology and Neurosurgery, London, UK) between 1989 and 2003. The principal aims of the study were to evaluate the factors guiding the decision to proceed to brain biopsy in patients with dementia, the role of the procedure in arriving at a diagnosis and in guiding management, and the risks of the procedure.

Methods

Case ascertainment

The case notes of all adult patients undergoing cerebral biopsy for the investigation of dementia at the National Hospital for Neurology and Neurosurgery between December 1989 and December 2003 were reviewed. In each case, at least two consultant neurologists had been involved in the decision to proceed to brain biopsy. For the purpose of this series, ‘dementia’ was defined as a progressive decline in cognitive function, behaviour or personality constituting the most significant component of the illness. Patients with acute encephalitides, intracranial tumours or epilepsy were not included in the series. In order to characterize the clinical population from which brain biopsy cases were drawn, the diagnostic profile of all patients referred to the dementia service of the National Hospital during the study period was estimated from clinical databases.

The biopsy procedure

Open brain biopsies were performed by the same neurosurgical team using a standard operative procedure. The procedure was performed under general anaesthesia. The skin was prepared and draped in a routine fashion. From 2001, disposable surgical instruments were used routinely. A burrhole was cut and enlarged with bone rongeurs to expose the dura over a clear 12–14 mm2 area. The dura was opened and excised and a 10 mm cube of underlying leptomeninges, grey and white matter was removed en bloc using sharp dissection. The specimens included meninges and grey and white matter. After removal of the specimen, haemostasis was achieved using diathermy and onlay of absorbable haemostatic gauze. The surgical wound was closed routinely with no drain and a dressing applied.

Examination of biopsy material

All biopsy specimens had been reviewed by experienced neuropathologists following standard procedures at the time of diagnosis. No further studies were undertaken at the time of this review, in order to define more accurately the diagnostic value of standard neuropathological methods in the examination of brain biopsies for dementia. The adequacy of the biopsy specimen was assessed in each case at the time of the original diagnosis; criteria for adequacy were a full thickness, 1 cm3 sample including cortex, underlying white matter and overlying leptomeninges. All specimens were fixed in 10% formalin, processed in paraffin wax and sectioned using routine procedures. Specimens were examined using standard histopathological stains (including haematoxylin and eosin, Luxol fast blue/cresyl violet, periodic acid–Schiff, Glees or Bielschowsky silver stains). Additional immunohistochemical stains for the diagnosis of neurodegenerative disorders included glial fibrillary acidic protein (GFAP) (Dako, UK) for the assessment of gliosis, and tau (Dako), neurofilament cocktail (70 and 200 kDa; Eurodiagnostica, UK), phosphorylated neurofilament (2F11; Dako), amyloid-β peptide (DAKO), prion protein (34F, KG9 and 6H4; Dako), ubiquitin (Dako) and α-synuclein (Institute of Psychiatry, London, UK). Immunological stains for inflammatory disorders included B-cell marker CD20 (Dako), T-cell markers CD3 (Dako) and CD68 (Dako), performed using standard avidin–biotin immunohistochemical methods. In situ hybridization of JC virus was performed when progressive multifocal leucoencephalopathy was suspected.

Case analysis

Each case was reviewed using a standard protocol (Table 2). Clinical details, laboratory investigations, brain MRI, EEG, CSF and brain biopsy findings were recorded. The reasons for biopsy, treatment before and after biopsy, and biopsy-related complications were also recorded in order to assess the impact of brain biopsy on management and the risks of the procedure. Case data were coded (Table 2) and univariate and multivariate (reverse stepwise) logistic regression analyses were performed using a standard statistical package (Stata8©, Stata Corporation, TX, USA) to determine which clinical, imaging and laboratory parameters or combinations of parameters predict findings at brain biopsy. Significance levels were assessed using the likelihood ratio test.

View this table:
Table 2

Information recorded for cases in this series

Information categoryCoding criterionNo. by category% of total*MeanRange
Clinical details
    GenderMale/female50/4055/45
    Age at biopsyAge (years)50.516–74
    Illness durationNo. of months28.41.5–144
    Severity of cognitive impairmentMild7 (63)11
Moderate12 (63)19
Severe44 (63)70
Clinical features
    SeizuresPresent1921
    Behavioural changesPresent6673
    HallucinationsPresent1618
    MyoclonusPresent3539
    Primitive reflexesPresent2022
    Other neurological signsPresent7786
Systemic featuresPresent2831
Systemic investigations
    Erythrocyte sedimentation rateValue (mm/h)211–140
    Raised: >2023 (74)31
    Raised: >507 (74)9
Other blood screens; chest X-ray, other imaging CSFAbnormal1517
    White blood cell countCells present:37 (84)44
value (no./µl)51–81
    Raised: >315 (84)18
    Raised: >57 (84)8
    Raised: >101 (84)1
    Protein concentrationValue (g/l)0.640.1–3.7
Raised: >0.722 (79)28
    Oligoclonal bandsPositive31 (81)38
Matched15 (81)19
Unmatched16 (81)20
    14-3-3Positive10§ (34)29
    S100 concentrationValue (ng/ml)0.580.08–2.7
    Raised: >0.3816 (32)50
EEG
    Alpha rhythmAbsent42 (87)48
    Epileptiform changesPresent12 (87)14
    Periodic abnormalitiesPresent4** (87)5
Brain MRI
    AtrophyPresent52 (86)60
Generalized36 (86)42
Focal16 (86)19
    White matter signalAbnormalities53 (86)62
Other brain imaging
    Cerebral angiographyAbnormal0 (7)
    Brain SPECT or PETAbnormal5 (5)100
Brain biopsy
    Indications, technical details, complications, histopathological findingsDetailsSee Tables 1 and 3
Treatment pre-/post-brain biopsyDetails
Alteration in management based on biopsySpecific therapy1011
Symptomatic therapy (AchEI)56
Other tissue biopsy††
    MuscleAbnormal3 (12)25
    Small bowel1 (8)13
    Skin (axillary)0 (5)
    Liver0 (4)
    Bone marrow0 (4)
    Temporal artery0 (2)
    Salivary gland0 (1)
    Tonsil0 (1)
    Vitreous0 (1)
  • The total number of cases in each category is indicated in parentheses where information is not available or not applicable in all cases. AchEI = acetylcholinesterase inhibition.

  • * Total number of cases for which data are available for that category;

  • based on Mini-Mental State Examination Score, Clinical Dementia Rating (where available), level of dependency and clinical impression as recorded in case notes;

  • includes cases with positive auto-antibody screens, abnormal liver function tests, raised serum ferritin, raised serum lactate, abnormal gallium scan;

  • § includes trace-positive cases;

  • ** three cases of CJD;

  • †† prior to brain biopsy (more than one tissue type in some cases).

Results

Case characteristics

A total of 90 cases were included. Case characteristics that were coded in the analysis of the series are summarized in Table 2. Fifty cases (55%) were males. The mean age at biopsy was 50.5 years (range 16–74 years) and the mean duration of symptoms prior to biopsy was 28 months (range 1.5–144 months). In all cases, brain biopsy was undertaken to exclude a reversible (inflammatory or infectious) process, suspected on the basis of features suggestive of CNS inflammation or other features that were considered atypical for a primary neurodegenerative disorder. The level of cognitive impairment at the time of biopsy was classified as severe in the majority of cases where this could be reliably assessed based on indices such as the Mini-Mental State Examination score or Clinical Dementia Rating (44/63 cases; 70%). The majority of patients coming to biopsy had neurological signs (82 cases; 91%) and/or behavioural abnormalities (66 cases; 73%); the most frequent findings on investigation were an absence of α rhythm on EEG (42/87 cases; 48%), and the presence of atrophy (52/86 cases; 60%) and/or abnormal white matter signal on brain MRI (53/86 cases; 62%). Among all patients referred to the dementia service during the study period who received a clinical diagnosis of dementia (approximately 1750 patients; Table 3), the leading diagnostic categories were Alzheimer's disease or mixed Alzheimer-vascular dementia (47%), frontotemporal lobar degeneration (22%), and other neurodegenerative conditions (12%): this diagnostic profile is similar to that projected for the major neurodegenerative causes of young onset dementia in the UK population (Harvey et al., 2003).

View this table:
Table 3

Summary of histopathological findings at brain biopsy in this series

FindingsBrain biopsiesAll referrals
No.%No.%
Diagnostic biopsiesAlzheimer's disease161882047
CJD: classical/variant10/112201
Pick's disease44380§22
Other neurodegenerations:3321012
    DLB, CBD, other tauopathy
Chronic (meningo-)encephalitis†44
Other inflammatory processes*:44
    vasculitis, Behçet's disease,
    neurosarcoidosis, granulomatous encephalopathy
Multiple sclerosis22
Vasculopathies*:441006
    atherosclerosis, congophilic angiopathy, CADASIL,
    Spatz–Lindenberg disease
Infections*:22
    Whipple's disease, PML
    Paraneoplastic encephalopathy11
Subtotal5157
Non-diagnostic biopsiesNon-specific gliosis3337
Gliosis, atypical PrP staining44
Gliosis, paraneoplastic encephalitis11
Normal11
Subtotal3943
Total901001750100
  • Estimated numbers and frequencies of clinical diagnoses (where available) among all patients referred to the dementia service during the study period are shown in italics. CADASIL = cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy; CBD = corticobasal degeneration; CJD = Creutzfeldt–Jakob disease; DLB = dementia with Lewy bodies; PML = progressive multifocal leucoencephalopathy; PrP = prion protein.

  • * Single cases within diagnostic grouping;

  • treatable process;

  • includes cases diagnosed with mixed Alzheimer–vascular dementia;

  • § all cases with clinical diagnosis of frontotemporal lobar degeneration.

Technical details and complications

Open biopsy from the non-dominant (right) frontal lobe was performed in 81 cases (90%). In the remaining nine cases, an alternative biopsy site was determined by the location of focal abnormalities: these other sites comprised the left frontal lobe (three cases), right temporal lobe, left occipital lobe (two cases at each site), right parietal lobe and right occipital lobe (one case at each site). Specimens were assessed as technically inadequate (small sample size) in five cases of the entire series (6%). Biopsy-related complications occurred in 10 cases (11%): these comprised generalized seizures (three cases), intracerebral haemorrhage (two cases), wound infection (two cases, both clinically minor), cerebral abscess (one case), subdural haematoma (one case) and perioperative pneumonia (one case). In all cases, complications responded to appropriate therapy without lasting neurological sequelae. Factors that may have contributed to the development of complications were identified in two cases (postoperative immunosuppression associated with cerebral abscess, and the use of an intraoperative biopsy catheter associated with intracerebral haemorrhage). There were no biopsy-related deaths in this series.

Histopathological findings

Histopathological findings are summarized in Table 3. Brain biopsy supplied a specific diagnosis in 51 cases (57%) from the entire series; a specific diagnosis was obtained from the majority (8/9 cases) of biopsies directed by focal abnormalities, and from a minority (1/5 cases) of technically inadequate biopsies. The most frequent diagnoses for the entire series were Alzheimer's disease (16 cases; 18%) and CJD (10 cases of classical CJD, one case of variant CJD; 12%). A potentially reversible process was identified in nine cases (10%): these comprised eight cases with inflammatory features and one case of Whipple's disease. The largest proportion of the non-diagnostic group and the largest single category within the entire series (33 cases; 37%) had non-specific gliosis variably affecting cortex and white matter, without distinctive histopathological features. Four cases classified within the non-diagnostic group had features suggestive of CJD; however, staining for prion protein was considered atypical.

Predictors of tissue pathology

Table 4 summarizes clinical and laboratory features that were identified as predictors of tissue pathology at brain biopsy. In a univariate analysis (threshold P < 0.05), an increased CSF cell count was the only significant predictor of an inflammatory (potentially reversible) process. However, raised cell count alone had limited positive predictive value and sensitivity: for example, an inflammatory process was present at biopsy in only 3/7 cases (43%) with CSF cell count >5 cells/µl, while a cell count >5 cells/µl was present in only 3/7 cases with inflammatory biopsies where CSF results were also available. The presence of primitive reflexes was a univariate predictor of Pick's disease at biopsy. Rapid clinical evolution, myoclonus and periodic complexes on EEG were each univariate predictors of a biopsy diagnosis of CJD. Raised CSF protein and the presence of matched oligoclonal bands in CSF and serum were each univariate predictors of a biopsy finding of non-specific gliosis. In a multivariate analysis (reverse stepwise regression; threshold P < 0.05), shorter illness duration and periodic complexes on EEG were independent predictors of CJD, and the constellation of behavioural change, raised CSF protein and matched oligoclonal bands in CSF and serum was associated with a finding of non-specific gliosis.

View this table:
Table 4

Predictive value of clinical and laboratory features for pathological findings at brain biopsy in this series

Pathological findingShorter illness durationMyoclonusPrimitive reflexesBehavioural changeEEG: periodic complexesCSF
Raised cell countRaised protein concentrationMatched OCBs*
Pick's diseaseU
Creutzfeldt–Jakob diseaseU, MUU, M
Inflammatory disorderU
Non-specific gliosisMU, MU, M
  • Empty cells indicate failure to reach significance (P < 0.05) in univariate or multivariate analyses. OCBs = oligoclonal bands; U = predictor in univariate analysis (P < 0.05); M = independent predictor in multivariate (reverse stepwise regression) analysis (P < 0.05).

  • * Matched oligoclonal bands in CSF and serum;

  • see Table 3 for diseases in this category.

No other statistically significant associations were identified in this series. Non-discriminatory factors at the prescribed significance threshold (P < 0.05) included: age, focal neurological signs, seizures, or clinical features of systemic disturbance; raised ESR; the presence of focal atrophy or white matter signal abnormalities on brain MRI; abnormalities of α rhythm or epileptiform discharges on EEG; unmatched oligoclonal bands and the concentration of S100 or 14-3-3 in CSF. Numbers in this series were insufficient to assess the predictive value of cerebral angiography, metabolic brain imaging or biopsy of tissues other than brain.

Outcome of brain biopsy

Treatment was directly determined by findings at brain biopsy in 10 cases (11%). Based on biopsy findings, immunosuppressive therapy was introduced in four cases, intensified (addition of cyclophosphamide to steroid therapy) in three cases, continued in one case, and discontinued in one case. Antibiotic therapy was introduced in the single case of Whipple's disease. Symptomatic therapy (acetylcholinesterase inhibition) was introduced in a further five cases (6%) following a biopsy diagnosis of Alzheimer's disease. It was not possible retrospectively to quantify the value of brain biopsy in terms of reduced diagnostic uncertainty and the provision of information to patients' relatives.

Information regarding long-term outcome after brain biopsy is incomplete. One patient died in hospital 2 weeks following the procedure, as a consequence of the underlying disease (biopsy in this case showed non-specific gliosis). A number of patients were transferred to the care of regional centres and lost to follow-up. Of those followed at the National Hospital (chiefly those with a neurodegenerative process or non-specific biopsy findings), the majority showed steady progression of their neurological deficit. Thirteen patients with neurodegenerative features at biopsy were known to have entered institutional care at the time of last contact. Two patients were known to have improved substantially following the introduction of immunosuppressive or antibiotic therapy based on biopsy findings, while several patients with non-diagnostic biopsy findings appeared to improve either spontaneously or with empirical immunosuppression. Ten patients in the series came to post-mortem: post-mortem examination corroborated the brain biopsy diagnosis in four patients (two cases of Alzheimer's disease, one case of Pick's disease and one case of CJD). Post-mortem supplied a specific diagnosis in five patients with non-diagnostic brain biopsies: one case with occasional vacuoles and atypical prion protein staining at biopsy was diagnosed as CJD; three cases with non-specific gliosis at biopsy were diagnosed as CJD, multiple sclerosis and cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL), respectively; while the remaining case had non-specific gliosis at biopsy and findings suggestive of mitochondrial cytopathy at post mortem.

Trends over time: diagnostic yield and complications

The number of biopsies, both diagnostic and non-diagnostic, and biopsy-related complications for each year of the series are displayed in Fig. 1. Over this 14-year period the annual number of brain biopsies showed a net increase; however, no trends were evident in the proportion of diagnostic biopsies or the incidence of complications associated with the procedure.

Fig. 1

Total number of brain biopsies for dementia (filled squares), diagnostic brain biopsies (open squares) and biopsy-related complications (open triangles) for each year of this series.

Discussion

The role of brain biopsy in dementia

In a proportion of patients with dementia the identification of a treatable cause may depend on a tissue diagnosis, yet limited evidence exists to guide the clinician in the difficult decision to proceed to brain biopsy. Here we have addressed this issue in a detailed clinical, laboratory and histopathological retrospective analysis of a large consecutive case series of brain biopsies for the diagnosis of dementia. Our findings demonstrate that a specific diagnosis can be established in over 50% of cases of dementia coming to brain biopsy, and a treatable disease can be identified in approximately 10%. Brain biopsy is not an innocuous procedure, even in experienced hands: 11% of patients in this series suffered significant complications, albeit without lasting sequelae. The diagnostic yield and complication rate of brain biopsy remained essentially constant over the course of this series (Fig. 1) and are compatible with the evidence of previous brain biopsy series in dementia spanning the last half-century (Table 1). In principle, the constant diagnostic yield of brain biopsy over time might reflect two opposing trends: on the one hand, improved sensitivity and specificity of histopathological methods, and on the other, improved techniques for non-invasive diagnosis with an increased proportion of atypical cases coming to biopsy. However, a relatively small number of tissue diagnoses (chiefly Alzheimer's disease, prion diseases and other neurodegenerative diseases) account for a similar proportion of cases across published brain biopsy series (Table 1).

The evidence from this and previous studies suggests that the risk: benefit analysis for brain biopsy in dementia is finely balanced between the potential for guiding the institution of specific therapy and the risk of complications associated with the procedure. However, such an analysis does not capture a number of important additional benefits that are more difficult to quantify. Brain biopsy may resolve diagnostic uncertainty (for example, by excluding a treatable process) and facilitate informed counselling of the patient's family. The value of brain biopsy would be substantially increased should disease-modifying therapies become available for neurodegenerative disorders such as Alzheimer's disease: this is now a realistic prospect for the first time since the procedure was introduced (Hock et al., 2003). While much interest currently centres on the development of reliable surrogate markers for tissue pathology in the common dementias, a proportion of cases will continue to have atypical findings; in such patients, tissue diagnosis is likely to remain an important clinical issue.

The relative benefit of brain biopsy would be enhanced if case selection were based on reliable and clinically useful predictors of specific biopsy findings. Our analysis of this large series of clinically atypical cases suggests that such predictors can be identified: we found that a raised CSF cell count predicts the finding of an inflammatory (potentially reversible) process at biopsy, whereas other clinical and laboratory features are associated with degenerative dementias for which no specific therapy is presently available. However, the positive predictive value of clinical and laboratory parameters alone would not be sufficient to guide clinical decision-making in the absence of a tissue diagnosis. For example, in this series, four cases (57%) with a CSF cell count in excess of 5/µl did not have biopsy confirmation of an inflammatory process. Conversely, features that predicted a degenerative process were also present in cases with inflammatory findings at brain biopsy. These parameters should not therefore replace brain biopsy: rather, they could be used in selecting patients for brain biopsy, which then offers the best chance of a definitive diagnosis.

The significance of the non-diagnostic brain biopsy: non-specific gliosis

Most published series of brain biopsy in dementia include a substantial proportion of specimens with non-specific histopathological findings such as cortical and white matter gliosis (Table 1). The nature of the underlying disease mechanism remains obscure in these cases despite the evolution of immunohistochemical and other contemporary neuropathological techniques. The commonest single pathological finding in this and several previous series was diffuse gliosis, variably affecting cortex and subcortical white matter. Progressive subcortical gliosis was first described as a sporadic condition by Neumann and Cohn (1967), and linkage to chromosome 17 and tau mutations has been established in several families (Lanska et al., 1994; Petersen et al., 1995). While this raises the possibility of a common mechanism linked to protein tau dysfunction in sporadic cases with similar histology, to date no specific cellular inclusions have been identified.

In the absence of a distinctive histopathological signature, it might be that non-specific gliosis is the result of a final common pathogenetic pathway triggered by an undefined spectrum of insults. In the present series, the finding of non-specific gliosis at brain biopsy was associated with a constellation of clinical and laboratory features (prominent behavioural change clinically, with raised CSF protein and matched oligoclonal bands in CSF and serum; Table 4): this evidence suggests that non-specific gliosis may have a uniform phenotype which remains to be defined fully in larger series and with post-mortem examination. It follows that a specific, as yet unidentified pathogenetic mechanism may be responsible. It is unlikely that such cases simply represent a more advanced stage of an unrecognized inflammatory process, since no common clinical or laboratory predictors of both non-specific gliosis and inflammatory change were identified. The relative prominence of behavioural abnormalities is consistent with early disruption of frontosubcortical mechanisms by a diffuse pathological process, while the finding of matched oligoclonal bands in CSF and serum is compatible with an autoimmune insult that produces a systemic immune response. This putative immune response could be instigated by a number of different processes. One potentially relevant class of disorders, the autoimmune channelopathies, is represented by the recently described encephalopathy associated with antibodies directed against CNS voltage-gated potassium channels (Vincent et al., 2004). One patient in the series of Vincent and colleagues (Vincent et al., 2004) underwent brain biopsy: diffuse gliosis was present; however, this was accompanied by lymphocytic infiltrates. We also note certain phenotypic and pathological similarities between our cases with non-specific gliosis and cases of encephalitis lethargica-like syndrome, another entity for which an autoimmune basis has been proposed (Dale et al., 2004).

Unresolved issues

The retrospective nature of the present study imposes several limitations. The identification of clinical and laboratory correlates of tissue pathology was subject to information recorded to the time of biopsy: it was therefore not possible to assess these associations systematically. In particular, MRI scans were not re-evaluated in the light of the biopsy findings: it is possible that the apparent lack of predictive value of white matter signal change reflects the inclusion of cases with typical vascular changes as well as atypical patterns of signal change. In this series, however, our objective was to identify predictors based on current diagnostic techniques and criteria that are part of the standard evaluation of patients with dementia. Previous authors have stressed the importance of correlating brain biopsy findings with the results of post-mortem examination, particularly where biopsy has been inconclusive (Hulette et al., 1992). In the present series, post-mortem examination either corroborated or advanced the results of brain biopsy; however, the number of patients coming to post-mortem was too small to allow a reliable correlation between post- and ante-mortem findings. These limitations should be addressed in a prospective analysis. It remains unclear whether the diagnostic yield of brain biopsy might be improved by the use of ancillary investigations such as structural or metabolic brain imaging to guide the location of biopsy in cases where the disease process is focal or inhomogeneous; it is likely, however, that the risk of exacerbating cognitive or neurological deficit will continue to dictate a biopsy site within the non-dominant frontal lobe in the majority of cases. A potentially more important issue concerns the appropriate timing of brain biopsy in the course of the illness. The majority of patients in the present series came to brain biopsy at a stage when cognitive impairment was already severe. It is possible that specific histopathological or immunochemical signatures may have been evident transiently at an earlier stage of the disease in those cases with a biopsy finding of non-specific gliosis.

Conclusions

Brain biopsy has an important role to play in the investigation of patients with dementia where a specific diagnosis cannot be made by standard non-invasive means. The present series suggests that certain clinical and laboratory features increase the likelihood of identifying a cause for dementia at biopsy, and the finding of a raised CSF cell count is the best single marker for a potentially treatable, inflammatory process. However, there remains a significant proportion of patients with non-specific findings (diffuse gliosis) even with the application of contemporary methods of tissue examination. Such cases may constitute a distinct clinicopathological entity that remains incompletely defined. Due to its retrospective nature, this study should be regarded as an exploratory analysis and our findings should be confirmed in prospective studies. Additional unresolved issues that should be addressed prospectively include the appropriate timing of brain biopsy in the course of dementia, the long-term outlook of patients undergoing brain biopsy, and validation of the procedure with pathological findings post-mortem. The value of brain biopsy in the management of dementia is likely to increase as disease-modifying therapies for neurodegenerative conditions become available.

Acknowledgments

The authors thank Dr Chris Frost and Dr Jane Warren for statistical advice and Ms Suzie Barker for assistance in analysis of clinical records. J.D.W. is supported by an EC Grant to the APOPIS (Abnormal Proteins in the Pathogenesis of Neurodegenerative Disorders) Consortium. J.M.S. is an Alzheimer's Society (UK) Clinical Research Fellow. N.C.F. and M.N.R. are supported by the UK Medical Research Council. J.H. receives support from the Reta Lila Weston Institute.

References

View Abstract