Subdural Spread of Local Anesthetic Agent following Thoracic Paravertebral Block and Cannulation

A 49-yr-old woman with a lung bronchogenic carcinoma was scheduled for a medial lobectomy. Her medical history included asthma treated with inhaled bronchodilators; there was no history of medication allergies.

No premedication was given. In the operating room, a 16-gauge intravenous access was obtained and the usual monitoring (electrocardiography, pulse oximetry, and noninvasive blood pressure monitoring) was initiated without any remarkable findings. Five hundred milliliters of Ringer's lactate solution was infused, and induction of anesthesia was initiated with 120 mg propofol, 150 μg fentanyl, and 50 mg rocuronium. A double-lumen tube was then inserted (Robertshaw, left side, no. 39). Anesthesia was maintained with a 0.8–1.2% concentration of sevoflurane and a rocuronium perfusion (0.6 mg · kg⁻¹· h⁻¹), and her lungs were ventilated with a fraction of inspired oxygen of 1.

After induction of anesthesia, the patient was placed on her left side in order to perform a right paravertebral block with cannulation for management of the perioperative pain (surgery and postoperative pain). An 18-gauge Tuohy needle (Perisafe® Plus; Becton-Dickinson, Warwickshire, Switzerland) was inserted 2.5 cm lateral to the aspect of the spinous process of T6 and advanced perpendicular to the skin in all planes. After encountering the costotransverse ligament, we advanced the needle approximately 1.5 cm further until there was loss of resistance to the injection air.

After negative aspiration, 3 cm of the epidural catheter (closed-end round catheter, 20 gauge) was inserted into the paravertebral space. The catheter was then checked with the negative aspiration test, and 3 ml bupivacaine, 0.5%, with epinephrine (1:200,000) was injected. There were no hemodynamic changes during the next 10 min, and another 12 ml of bupivacaine, 0.5%, without epinephrine was injected. Fifteen minutes after this second bolus, the patient's blood pressure decreased substantially (to 70/40 mmHg), as did the heart rate (48 beats/min). There were no wide variations on the arterial pressure curve area related to mechanical ventilation. All lung auscultation findings, airway pressures, and pressure–volume curves were absolutely normal during the period between intubation and the beginning of one-lung ventilation.

Sevoflurane administration was discontinued, followed by the rapid infusion of 500 ml of a colloid and the administration of 10 mg ephedrine intravenously. The dosing of ephedrine was repeated twice but did not increase the systolic blood pressure to higher than 100 mmHg, so a 0.1-μg · kg⁻¹· min⁻¹infusion of norepinephrine was initiated, with good results. During the next 60 min the norepinephrine infusion rate was progressively reduced, and the infusion was finally stopped due to hemodynamic stabilization. The thoracotomy was started in this left lateral position, and sevoflurane administration was reinitiated then. The heart rate did not change for the rest of the surgery (180 min total). No other analgesic administration was needed during the rest of the surgery until the thoracic wall closure, when a slight increase in the heart rate and blood pressure suggested some pain, and 100 μg fentanyl was given intravenously.

At the end of surgery (40 min after this fentanyl administration and 10 min after the end of sevoflurane administration) the patient was awake and helpful, and the ventilation pattern was good, so she was extubated. She was moved to the intensive care unit. The patient did not have any sensory or motor blockade. Once the patient had been monitored and no hemodynamic, ventilatory, or any other changes were noted, we decided to check the placement of the catheter, with the patient's permission. After a negative aspiration test, 15 ml of the contrast medium Iopamiro-300 (Iopamidol, 300 mg/ml; Rovi, Madrid, Spain) was injected through the catheter with the patient seated. Two chest radiographs were obtained: lateral (fig. 1) and posterior–anterior. The lateral view (fig. 1) showed the characteristic “railroad tracks” from the lumbar region to the cervical region. Furthermore, computed tomography thoracic–abdominal imaging (fig. 2) was performed to confirm the contrast extension. The contrast circumferentially outlined the thecal sac (subdural spread sign).

The catheter was removed immediately. A postoperative intravenous analgesia infusion (meperidine and ketorolac) was initiated and was continued for the next several days in the wards. The patient was discharged from the wards 7 days later, with no other incidents during her hospitalization.

Subdural cannulation has been described as a complication of intended anesthetic blocks (epidural, dural, and even interscalene block), but there are no published reports of a subdural block with an intended paravertebral block. However, some cases of total spinal block 1have been described, and a postpuncture headache has been reported as a possible complication of paravertebral block. 2 

The subdural space is a potential space located between the dura mater and the arachnoid mater. The dura mater extends laterally some “cuffs” along the nerve roots at the spinal level, which are in contact with the spinal ganglion. Sometimes these dural cuffs are prolonged to the paravertebral space as a normal anatomic variation. Abnormal variations (meningocele, perineural cysts) are possible and can be discovered in image studies with contrast medium. The radiologic studies of our patient could not completely eliminate the possible existence of these abnormal anatomic variations that would have facilitated the subdural block. Another possibility could be that the needle advancement was anterior–medial instead of directly anterior, which would greatly increase the chances of a dural cuff injury and/or entry in the subdural space.

Lubenow et al.  3described the two major and the three minor clinical criteria for the diagnosis of subdural block. Our patient met one of the major criteria (negative aspiration), and the other—unsuspected widespread sensory block—was probably present but masked by the simultaneous general anesthesia. The minor criteria included delayed onset and massive sympatholysis (hypotension and possibly pupil dilatation), which was out of proportion to the dose of local anesthetic agent injected. The third minor criterion described by Lubenow et al.  3, which may have been present but masked by the general anesthesia, was variable motor block. This last minor criterion is an issue with intended epidural anesthesia but is not a consideration with intended thoracic paravertebral block.

The contrast medium injected through the paraspinal catheter surrounded the spinal cord except for the site of the spinal ganglia, a circumstance that was related to the extraarachnoid location of the contrast medium, but it was impossible to determine where the tip of the catheter was. The extension of the contrast medium along several vertebral bodies—near the complete dorsal and lumbar spine, as was shown in the lateral view of the chest radiograph—points out the contrast in the subdural space; if it had been in the epidural space, the contrast would have been limited to a smaller extension. 4 

The occurrence of hypotension with paravertebral blocks usually is related to hypovolemia by unilateral vasodilation, 5but the paravertebral volume injected in this patient was too small (15 ml) to produce an extended sympathetic block, which could have explained the profound hypotension. Recently, Saito et al.  6showed an increase in blood pressure when 22 ml lidocaine was injected into the paravertebral space. In our patient, the extreme hypotension was the clue that made us suspect a block other than a paravertebral one.

Four different blocks had to be considered: thoracic paravertebral block, thoracic epidural block, subarachnoid block, and subdural block. Clinical manifestations of a subdural block at the thoracic level range from hypesthesia of the affected dermatomes to full respiratory arrest with severe hypotension or cardiac arrest. 3Definitive diagnosis requires radiologic studies with contrast medium injected through the catheter.

The low incidence of severe complications associated with paravertebral blocks makes anesthesiologists confident about this technique. A good knowledge of the anatomy of the paravertebral space is important to understanding possible severe complications. The number of paravertebral catheters being placed for management of postoperative pain is great, but the incidence of complications, such as malpositioning of catheters, is unknown. Abnormal behavior of local anesthetics administered in a paravertebral catheter should lead one to suspect a subdural block; its early detection can prevent serious consequences. The removal of these catheters is necessary due to the unpredictable behavior of anesthetics in the subdural space.