Show Summary Details
Page of

Lumbar puncture 

Lumbar puncture
Lumbar puncture

Roger A. Barker

, Wendy Phillips

, and R. Rhys Davies

Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy).

Subscriber: null; date: 11 December 2017


Lumbar puncture provides the means to sample cerebrospinal fluid for diagnostic purposes and to remove it for some therapeutic purposes. The procedure allows measurement of the pressure of cerebrospinal fluid, its cytological composition, biochemical content, and microbial as well as serological characteristics.

Indications—the commonest diagnostic indications are clinical suspicion of central nervous system infection (meningitis, encephalitis), subarachnoid haemorrhage, and demyelinating diseases (central and peripheral); the commonest therapeutic indications are idiopathic intracranial hypertension and for intrathecal administration of drugs.

Containdications—these include infection in the skin overlyingthe spine, evidence of intracranial hypertension, and bleeding disorders.

Acknowledgement: This chapter has been adapted from the one contributed to the 4th edition of the Oxford Textbook of Medicine by R A Fishman.


Lumbar puncture (LP) should be carried out only after clinical evaluation of the patient, with consideration of the potential value and hazards of the procedure. Cerebrospinal fluid (CSF) findings are important in the differential diagnosis of a range of central nervous system (CNS) infections (meningitis, and encephalitis), subarachnoid haemorrhage, confusional states, acute stroke, status epilepticus, meningeal malignancies, demyelinating diseases (central and peripheral), vasculitis, cerebral venous thrombosis, idiopathic intracranial hypertension, and normal pressure hydrocephalus.

Examination of CSF is usually necessary in patients with suspected intracranial bleeding when initial CT/MRI has not been diagnostic. LP is also helpful in patients with symptoms of raised intracranial pressure and papilloedema with normal imaging. Measurement of the opening pressure and examination of the CSF cellular constituents may distinguish idiopathic intracranial hypertension (IIH), venous sinus thrombosis, and diffuse meningitic processes such as neoplastic or inflammatory/infective meningitis.

LP is necessary in patients with suspected infection. The cellular constituents, as well as the glucose and protein concentrations, may suggest a type of organism. This may be confirmed by culture, or by using specific stains or polymerase chain reaction (PCR) for known CNS pathogens. Finally, CSF analysis is useful in the diagnosis of inflammatory disorders. Again, cellular constituents of the CSF may help to define the condition: raised protein in the absence of cells occurs with inflammatory polyneuropathies, and the detection of intrathecal immunoglobulin synthesis (i.e. oligoclonal bands on electrophoresis) may also be helpful.

LP has therapeutic as well as diagnostic uses. Drugs may be administered intrathecally in meningeal malignancies, fungal meningitis, spastic paraparesis, or spinal anaesthesia. Repeated LP (or placement of a lumbar drain) may also be used to reduce intracranial pressure in IIH, normal pressure hydrocephalus (NPH), or venous sinus thrombosis, or after surgical procedures (e.g. after acoustic neuroma excision).


LP is contraindicated in the presence of infection in the skin overlying the spine. A serious complication of LP is aggravation of pre-existing, unrecognized, brain herniation syndrome (e.g. uncal, cerebellar, or cingulate herniation) associated with intracranial hypertension. This hazard is the basis for considering papilloedema as a relative contraindication to LP. The availability of CT/MRI has simplified the management of patients with papilloedema. If CT reveals no evidence of a mass lesion, LP is usually needed in the presence of papilloedema to establish the diagnosis of IIH and to exclude meningeal inflammation or malignancy. In most acute hospital settings, CT is obtained before LP in the presence of new neurological signs or any features of raised intracranial pressure (including drowsiness).

Thrombocytopenia and other bleeding diatheses (including therapeutic anticoagulation) predispose patients to needle-induced subarachnoid, subdural, and epidural haemorrhage. LP should be undertaken only if urgently needed when the platelet count is depressed to about 50 000/μ‎l or below. Platelet transfusion just before the puncture is recommended if the count is below 20 000/μ‎l or dropping rapidly. The administration of protamine to patients on heparin, and vitamin K or fresh frozen plasma to those receiving warfarin, is recommended before LP to minimize the hazard of the procedure. In patients receiving subcutaneous low-molecular-weight heparin, standard practice is to delay puncture for at least 24 h after an injection.


Complications of LP include worsening of brain herniation and spinal cord compression, subarachnoid bleeding, diplopia, radicular symptoms, backache, and headache. Infection after an LP is rare. The introduction of dermal tissue into the subarachnoid would be a theoretical risk if an unstyletted needle were used.

Headache after LP is the most common complication, occurring in about 25% of patients and usually lasting 2 to 8 days. It results from low CSF pressure due to persistent fluid leakage through the dural hole. Characteristically, pain is present in the upright position and is promptly relieved by a supine position. Aching of the neck and low back is common. The headaches are aggravated by cough or strain and may be associated with nausea, vomiting, or tinnitus. They are less likely after first-pass LP with a small needle, and if the needle stylet is reinserted before removal. The management of headache after LP depends on strict bedrest in the horizontal position for at least 12 h, adequate hydration, and simple analgesics. Caffeine may help. If conservative measures fail, the use of a ‘blood patch’ is indicated. The technique consists of injecting autologous blood epidurally, close to the site of the dural puncture, forming a thrombotic tamponade that seals the dural hole. Blood patches are often undertaken by anaesthetists in their obstetric practice.

Cerebrospinal fluid

Opening pressure

The CSF pressure should be measured routinely. The pressure level within the right atrium is the reference level with the patient horizontal in the lateral decubitus position. The normal lumbar CSF pressure ranges between 50 and 200 mmH2O (and as high as 250 mmH2O in very obese individuals). Low pressures are seen in dehydration, spinal subarachnoid block, or CSF leak (e.g. from previous LP), or may be technical in origin because of faulty needle placement. Increased pressures occur with intracranial mass lesions (when an LP should not routinely be performed), infections, acute stroke, cerebral venous occlusions, congestive heart failure, pulmonary insufficiency, and IIH.

Leucocytes and cytology

Normal CSF contains no more than five lymphocytes or mononuclear cells per microlitre. A higher white cell count indicates disease in the CNS or meninges. A stained smear of the sediment is needed for an accurate differential cell count. A variety of centrifugal and sedimentation techniques has been used. A pleocytosis occurs in a range of neoplastic, infective, and inflammatory disorders, and the changes characteristic of the various meningitides are listed in Table 24.3.1. Other disorders associated with a pleocytosis include stroke, subarachnoid haemorrhage, cerebral vasculitis, acute demyelination, and brain tumours. CSF pleocytosis may occur in inflammatory polyneuropathies but is not characteristic, and should prompt a search for an underlying cause, e.g. HIV. Eosinophilia most often accompanies parasitic infections, such as cysticercosis. Cytological studies for malignant cells are rewarding in some CNS neoplasms although this may require repeated LPs (typically three) and the involvement of specialist haematology service for flow cytometry analysis to define immunophenotype. If malignancy is suspected, it is important to obtain a large volume (e.g. 10 ml) of fresh CSF for analysis.

Table 24.3.1 Cerebrospinal fluid findings in meningitis


Pressure (mmH2O)


Protein (g/l)

Glucose (mmol/l)

Acute bacterial

Usually elevated

Several hundred to more than 60 000; usually a few thousand but occasionally less than 100 (especially meningococcal or early in disease). Polymorphonuclears predominate

Usually 1 to 5, occasionally more than 10

0.2 to 2.2 in most cases (in the absence of hyperglycaemia)


Usually elevated; may be low with dynamic block in advanced stages

Usually 25 to 100; rarely more than 500. Lymphocytes predominate except in early stages when polymorphonuclears may account for 80 per cent of cells

Nearly always elevated, usually 1 to 2; may be much higher if dynamic block

Usually reduced; less than 2.5 in three-quarters of cases


Usually elevated

0 to 800; average 50. Lymphocytes predominate

Usually 0.2 to 5; average 1

Reduced in most cases; average 1.7 (in absence of hyperglycaemia)


Normal to moderately elevated

5 to a few hundred; but may be more than 1000, particularly with lymphocytic choriomeningitis. Lymphocytes predominate but there may be more than 80 per cent polymorphonuclears in the first few days

Frequently normal or slightly elevated; less than 1; may show greater elevation in severe cases

Normal (reduced in one-quarter of cases of mumps and herpes simplex)

Syphilitic (acute)

Usually elevated

Average 500. Usually lymphocytes; rarely polymorphonuclear

Average 1

Normal (rarely reduced)


Often increased; low with dynamic block

Increased mononuclears and polymorphonuclears with 2 to 7 per cent eosinophilia in about half of cases

Usually 0.5 to 2

Reduced in a fifth of cases


Normal to considerably elevated

0 to fewer than 100 mononuclear cells

Slight to moderate elevation

Reduced in half of cases


Normal or elevated

0 to several hundred mononuclears plus malignant cells

Elevated often to high levels

Normal or greatly reduced (low in three-quarters of carcinomatous meningitis cases)

Cerebrospinal fluid immunoglobulins are commonly increased in all of the above (including carcinomatous meningitis) as well as in multiple sclerosis and central nervous system vasculitis.

Cerebrospinal fluid immunoglobulins are assessed by the IgG index: (IgG (cerebrospinal fluid) × albumin (serum))/ (IgG serum × albumin(cerebrospinal fluid)). The normal index is less than 0.65.

Oligoclonal bands (with gel electrophoresis) present in cerebrospinal fluid but absent in serum are also a measure of abnormally increased cerebrospinal fluid immunoglobulins synthesized within the CNS).

Bloody CSF due to needle trauma contains increased numbers of white cells contributed by the blood. A useful approximation to a true white cell count can be obtained by the following correction for the presence of the added blood: if the patient has a normal full blood count, subtract from the total white cell count (WBC per μ‎l) one white cell for each 1000 red blood cells (RBCs) present. Thus, if bloody fluid contains 10 000 red cells and 100 white cells/μ‎l, 10 white cells would be accounted for by the added blood and the corrected leucocyte count would be 90/μ‎l. If the patient’s full blood count reveals significant anaemia or leucocytosis, the following formula may be used to determine more accurately the number of white cells (W) in the spinal fluid before the blood was added:


The presence of blood in the subarachnoid space produces a secondary inflammatory response, which leads to a disproportionate increase in the number of white cells. Following an acute subarachnoid haemorrhage, this elevation in the WBC is most marked about 48 h after onset, when meningeal signs are most striking.

To correct CSF protein values for the presence of added blood due to needle trauma, subtract 0.01 g for every 1000 RBCs. Thus, if the red cell count is 10 000/μ‎l and the total protein is 1.1 g/litre the corrected protein level would be about 1 g/litre. The corrections are reliable only if the cell count and total protein are both made on the same tube of fluid.


To differentiate between a traumatic spinal puncture and pre-existing subarachnoid haemorrhage (or subarachnoid extension of a parenchymal bleed), the fluid can be collected in a series of three tubes. In traumatic punctures, the fluid generally clears between the first and the third collections. This is detectable with the naked eye and may be confirmed by cell count. In subarachnoid bleeding, the blood is generally evenly admixed in the three tubes. A sample of the bloody fluid should be centrifuged and the supernatant fluid compared with tap water to exclude the presence of pigment. The supernatant fluid should be crystal clear if the red cell count is less than 100 000 cells/μ‎l but, with more severe bloody contamination, the plasma proteins may be sufficient to cause minimal xanthochromia.

Following subarachnoid haemorrhage, the supernatant fluid usually remains clear for 2 to 4 h, or even longer, after the onset of subarachnoid bleeding. Between 12 h and 12 days after symptom onset, however, the absence of xanthochromia effectively excludes the diagnosis. After an especially traumatic puncture, some blood and xanthochromia may be present for as long as 2 to 5 days after the initial puncture. CSF protein of greater than 1.5 g/litre, irrespective of cause, may be associated with faint xanthochromia. When the protein is elevated to much higher levels, as in spinal block, inflammatory demyelinating polyneuropathies, or meningitis, the xanthochromia may be marked. Xanthochromic fluid when the protein level is less than 1.5 g/litre generally indicates recent subarachnoid haemorrhage. Rarely, xanthochromia is due to severe jaundice, carotenaemia, or rifampicin therapy.

If there is any doubt as to whether the CSF reflects a subarachnoid haemorrhage as opposed to a bloody tap, further imaging of the vasculature is required.


Two major pigments derived from red cells, the basis of xanthochromia after haemorrhage, may be seen in CSF—oxyhaemoglobin and bilirubin. Methaemoglobin is detected only by spectrophotometry. Oxyhaemoglobin, released with lysis of red cells, may be detected in the supernatant fluid within 2 h of a subarachnoid haemorrhage. It reaches a maximum in about the first 36 h and gradually disappears over the next 7 to 10 days. Bilirubin is produced in vivo by leptomeningeal cells after red cell haemolysis, and is first detected about 10 h after the onset of subarachnoid bleeding. It reaches a maximum at 48 h and may persist for 2 to 4 weeks after extensive bleeding. The severity of the meningeal signs associated with subarachnoid bleeding correlates with the inflammatory response, i.e. the leucocytic pleocytosis.


Total protein

The total protein level of CSF ranges between 0.15 and 0.5 g/litre. although an elevated protein level lacks specificity, it is an index of neurological disease reflecting a pathological increase in the permeability of endothelial cells. Greatly increased protein levels, 5 g/litre and above, are seen in meningitis, bloody fluids, or cord tumour with spinal block. Guillain–Barré syndrome or chronic inflammatory demyelinating polyneuropathies (CIDPs), diabetic radiculoneuropathy, and myxoedema may also increase the level to 1 to 3 g/litre. Low protein levels, below 0.15 g/litre, occur most often with CSF leaks due to a previous LP or traumatic dural fistula.


A vast number of proteins may be measured in CSF but only increases in immunoglobulins are of major diagnostic importance. Such increases are indicative of an inflammatory response in the CNS and occur with immunological disorders, and bacterial, viral, spirochaetal, and fungal diseases. Immunoglobulin assays are most useful in the diagnosis of multiple sclerosis (MS), other demyelinating diseases, and CNS vasculitis. The CSF level is corrected for the entry of immunoglobulins from the serum by calculating the IgG index (see Table 24.3.1). More than one electrophoretic band in CSF (‘oligoclonal bands’) is also abnormal. This pattern is found in 90% of MS cases and may occur with other inflammatory processes (e.g. vasculitis and paraneoplasia). It is important to stress that, whenever a LP is performed and CSF sent, a serum sample is sent at the same time so that the necessary comparisons can be made to determine whether the inflammation is confined to the CNS (e.g. MS) or part of a multisystem disorder (e.g. systemic lupus erythematosus).

Other proteins

The 14-3-3 proteins are ubiquitous regulator proteins, and can be found in the CSF of patients with Creutzfeldt–Jakob disease. The astrocytic protein, S100β‎, and neuron-specific enolase may also be raised in the CSF of such patients. The sensitivity and specificity are limited. Angiotensin-converting enzyme levels are increased in neurosarcoidosis but variation between individuals undermines the value of this in diagnosis.


The concentration of glucose in CSF is dependent on the blood concentration. The normal range of glucose concentration in CSF is between 2.5 and 4.5 mmol/litre in patients with a blood glucose between 4 and 7 mmol/litre, i.e. 60 to 80% of the normal blood level. CSF glucose values between 2.2 and 2.5 mmol/litre are usually abnormal, and values below 2.2 mmol/litre invariably so. Hyperglycaemia during the 4 h before LP results in a parallel increase in CSF glucose. The latter approaches a maximum and the CSF:blood ratio may be as low as 0.35 in the presence of a greatly elevated blood glucose level and the absence of any neurological disease. Increased CSF glucose has no further diagnostic significance. The CSF glucose level is abnormally low (hypoglycorrhachia) in several diseases of the nervous system. It is characteristic of acute purulent meningitis, and is a usual finding in tuberculous and fungal meningitis, as well as neoplastic meningitis. It is usually normal in viral meningitis, although it can be reduced in patients with mumps, herpes simplex, and zoster meningoencephalitis. CSF glucose may be reduced in other inflammatory meningitides including cysticercosis, amoebic meningitis (Nagleria spp.), acute syphilitic meningitis, sarcoidosis, granulomatous arteritis, and other vasculitides. Glucose levels may also be depressed in the context of subarachnoid haemorrhage (usually 4 to 8 days after the bleed) or in chemical meningitis after intrathecal injection.

The major factor responsible for reduced glucose levels is increased anaerobic glycolysis in adjacent neural tissues and, to a lesser degree, in cells that may be detected in the CSF itself. As such, the decrease in the CSF glucose level is accompanied by an inverse increase in the CSF lactate level.


Detection of elevated CSF lactate may be useful when inborn errors of metabolism and mitochondrial disease are suspected. Although CSF lactate is more reliable than plasma lactate, it is not definitive and muscle biopsy, DNA analysis, etc. should also be performed. High CSF lactate is also seen in systemic lactic acidosis and nonspecifically when other CSF abnormalities are present. Some advocate the use of CSF lactate as a marker for bacterial versus viral meningitis, but this is not widely accepted.

Microbiological and serological reactions

The use of appropriate stains, cultures, and PCR to specific pathogens is essential in cases of suspected infection. Typically CSF is examined using standard stains (e.g. Gram stain) before culturing and PCR tests are undertaken. The involvement of an infectious disease specialist is often helpful in defining the true validity and significance of positive microbiological/serological tests. In all cases, if there is a clinical suspicion of a CNS infection, treatment should be commenced before any definite microbiological results from the CSF can be obtained.

Further reading

Fishman RA (1992). CSF in diseases of the nervous system, 2nd edition. W B Saunders, Philadelphia, PA.Find this resource: