top of page

Cerebrospinal Fluid Collection

Collection Techniques


CSF originates from production sites including the choroid plexus of each of the lateral, third, and fourth ventricles; the brain by way of the ependymal lining of the ventricular system and the pial-glial membrane covering its external surface; and by blood vessels in the pia-arachnoid.  The method of production is both by ultra-filtration from the blood plasma and by active transport mechanisms that utilize energy. 


Cerebrospinal fluid normally has low protein content and contains few cells.  Pathology in the CNS often is reflected in the cerebrospinal fluid when there is compromise of the blood-brain barrier, the blood-CSF barrier, or the CSF’s interface with the brain and spinal cord.  Abnormalities in the cerebrospinal fluid during disease are usually limited to changes in the number and distribution of cells present and increased protein content, making the analysis of cerebrospinal fluid a sensitive but rarely specific indicator of CNS disease. Because of this, the results of cerebrospinal fluid analysis must be interpreted in light of all other clinical findings and, ideally, advanced imaging.  Ancillary tests that can be performed on CSF to provide more specific results are covered in the final section.

In animals with lesions that localize to the brain or cervical spinal cord, there may be clinical benefit in collecting fluid from both the CM and lumbar cistern. In dogs with thoracolumbar myelopathy, CSF collected from the CM cistern may not be representative of the underlying disease process.

Collection Technique

General anesthesia is mandatory for CSF collection.  The site of the collection depends essentially on the site of the suspected pathology; as CSF flows predominantly in a cranial to caudal direction, abnormal CSF is more likely to be present caudal to a lesion. An exception to this generalization is in the case of cervical lesions, where a cisternal tap may be of higher yield than lumbar, because the proximity of the cerebellomedullary cistern allows affected CSF to reach it via bidirectional flow.  

Collection from the cerebellomedullary cistern (cisterna magna) is easier and less likely to produce a sample contaminated with blood.  Lumbar sampling is technically more difficult and more likely to produce blood contamination but is also more likely to be diagnostically useful for lesions in the thoracolumbar spine.


A minimum of 0.5ml (approximately 11 drops out of a 22 gauge needle) of CSF is needed for standard analysis by most laboratories.  One milliliter per 5kg of body weight can safely be removed at a time.  If CSF is to be sent for analysis to a distant laboratory, two aliquots should be obtained to allow for the preservation of one sample (see sample handling, below).

Cerebello-medullary cistern collection

The animal is anesthetized and a rigid endotracheal tube is used to avoid occlusion of oxygen flow during the neck flexion required for CSF collection from the cerebellomedullary cistern (CMC). The CMC is the most common site of collection in dogs and cats and is reached by inserting a spinal needle at the junction of the horizontal line between the occipital protuberance and the arch of the axis (C2 vertebrae) and the vertical line which runs along the cranial edges of the wings of the atlas (C1 vertebra). 

A 22 or 20-gauge, 1.5-inch spinal needle will usually reach the CMC in most dogs and cats, although in large dogs over 30kg a 2.5-inch needle may be needed.  

This region is clipped and aseptically prepared.

With the animal in lateral recumbency, an assistant holds the head in ventral flexion, so that the nose is parallel to the table while also ensuring that respiration is unencumbered. 

The animal has to be observed for adequate ventilation during the whole procedure, and positive-pressure ventilation is often required.  Adequate depth of anesthesia should be confirmed before beginning the procedure.  The needle is positioned directly on midline perpendicular to the neck and advanced slowly 1 to 2 mm at a time.  Once the skin has been penetrated, the stylette can be removed.  The needle is then slowly and carefully advanced. The puncture of the dura mater and atlanto-occipital membrane may be felt as a slight ‘pop’, but many times this is not the case.  Resistance will decrease when the dura is penetrated, and CSF flow is observed into the hub of the needle; the flow of CSF is the only reliable sign of a successful puncture.  The first two drops of CSF are allowed to drip freely to remove any debris collected during needle insertion, and the following drops are collected into a sterile collection tube. CSF should never be aspirated with a syringe, as this creates negative pressure in the subarachnoid space and increases the risk of herniation.  

If the needle hits bone while being advanced, it may be redirected cranially or caudally, moving the needle off the bone into the atlanto-occipital space.  If the nose is deviated from a perpendicular angle to the table, one of the vertebral sinuses will be entered and blood will be obtained instead of CSF.  Hitting a vertebral sinus will not prevent getting a clear CSF sample on a subsequent attempt, as these structures are extradural.  If blood-tinged CSF is observed initially, it may clear after several drops if it is due to injury to a subarachnoid vessel and can then be collected.  If the blood tinge does not clear, it is likely to be part of the disease process, and the CSF should be collected for analysis.  Once an adequate amount of CSF is collected, the needle can be slowly withdrawn.

Certain disease processes can obliterate the cerebellomedullary cistern or increase the viscosity of the CSF to the extent that it will not flow and may not even appear in the hub of the needle.  One must be aware of this possibility so as not to insert the needle too far while waiting for the flash of CSF to appear.  In cases where a CMC sample cannot be obtained, a lumbar puncture can be performed instead.

Lumbar Collection

Lumbar puncture is technically more difficult than CMC puncture and can be associated with more frequent blood contamination.  The site for lumbar puncture is L5-L6 (or L4-L5) in dogs and L6-L7 (or L5-L6) in cats.  The area is clipped and aseptically prepared.  The patient should be positioned in right lateral recumbency (can be in left lateral recumbency for a left-handed person), so that its dorsum is close and parallel to the edge of the table; the hind legs are flexed as much as possible.  The L6 spinous process is palpated just cranial to the wings of the ilium; it is much more prominent than the smaller L7 spinous process caudally.

The needle is inserted, with the stylet left in place, over the center of the spinous process of L6 if attempting to reach the L5-L6 space and advanced until it hits the spinous process.  The needle tip is then moved slightly laterally so that it can be ‘walked’ down the spinous process until it hits the dorsal lamina; it is subsequently advanced at a 45° angle with the needle pointed directed cranially remaining against the spinous process.  When correctly positioned, the needle typically passes through the interarcuate and yellow ligament; this feels as if the needle is going through rubber.  This step is performed “by feel” and multiple attempts may be necessary to achieve the correct angle of insertion.  If the needle hits bone while being advanced, it may be redirected cranially or caudally, moving the needle off the bone into interarcuate space while remaining alongside of the spinous process.  This process may lead to bending of the needle, so a 20g needle is often required in larger dogs (2.5” or 3.5”). If it is no longer possible to visualize where the tip of the needle is in relation to the anatomy, the needle should be withdrawn for a new attempt.  

The needle will pass through the yellow ligament and the cauda equina/caudal spinal cord.  The latter often results in a tail or leg twitch seen or felt by the holder.  The needle is advanced to the floor of the vertebral canal.  When the stylet is removed, CSF flow is observed, which is the only reliable sign of a successful puncture. CSF may be collected from either the dorsal or ventral lumbar subarachnoid space.  If CSF does not appear, the needle can be withdrawn slightly off of the floor of the canal or carefully turned while in place.

If blood appears, the needle should be withdrawn and a new attempt made.  If blood-tinged CSF appears, it may clear in a few drops.  If the blood tinge does not clear, it is likely part of the disease process, and the CSF should be collected.  Complications of passing the needle through the caudal spinal cord and nerve roots are rarely reported, but this procedure can cause intradural or intramedullary hemorrhage.  

Sample Handling

Cerebrospinal fluid should be collected into an empty sterile container.  If infection is suspected, CSF can be collected into culture media or a culturette swab can be used to take a sample out of a sterile container. 

If the analysis is going to be delayed by more than an hour, ideally at least two aliquots of CSF should be collected: one for analysis of protein, other analytes, and cell counts and another for cytologic analysis.  Because of the typically low protein content of CSF, the cells in a sample are labile and short-lived once collected, which can significantly alter the differential cell count and reduce the usefulness of the test.  Samples with high protein content (>50mg/dl) are unlikely to be affected by 8-12 hours of storage without a stabilizing agent, but because the protein content is often unknown at the time of collection, a stabilizing agent is recommended. Alternatively, slides can be prepared using one of the techniques below and mailed to a laboratory for staining and evaluation.

Autologous serum (added to achieve a concentration of 10%), hetastarch (added 1:1), and EDTA (added 1:1) are recommended as stabilizing agents to preserve cell integrity for up to 48 hours after collection.  Formalin is not recommended as a preservative if cytologic analysis is to be performed; added 1:1 it will suffice for cell count and protein analysis.  If too little CSF is collected to submit two separate aliquots, hetastarch should be used to preserve the sample; protein analysis will not be affected by the addition of hetastarch as it would by the addition of autologous serum or EDTA.

Samples should be clearly labeled when sent to the lab so that the dilutional effects of the stabilizing agent can be taken into account when necessary.  All samples should ideally be sent at a refrigerated temperature (4ºC).  

bottom of page