Indocyanine Green: Evidence of Neurotoxicity in Spinal Root Axons

Approval was obtained from the Stanford University Administrative Panel on Laboratory Animal Care. Sixteen male Sprague-Dawley rats weighing 320–420 g were anesthetized with enflurane (1–2.5%) and nitrous oxide (70%) in oxygen. As previously described, 21,22the spinal cord was exposed through a thoracolumbar laminectomy, and using microsurgical techniques, single lumbar DRs were excised and transferred to a perfusion and recording chamber. The perfusion compartment volume was approximately 1 ml, and the perfusate flow was typically 5 ml/min. The roots were continuously superfused with artificial CSF (aCSF; Na⁺, 150 mm; K⁺, 4 mm; Cl⁻, 127 mm; Ca²⁺, 2 mm; Mg²⁺, 1.3 mm; PO⁴⁻³, 1.2 mm; HCO³⁻, 26 mm; glucose, 11 mm) at 37°C and equilibrated with a 95% O²+ 5% CO²gas mixture to maintain a pH of 7.3–7.4. Single-fiber microdissection and recording techniques were used to isolate activity in individual spinal root axons. Dissection and recording were carried out on the proximal end of each DR after isolation from the main perfusion chamber.

Supramaximal (1.5 × threshold) constant voltage stimuli at 0.3 Hz (0.1-ms duration) were delivered to the distal end of the isolated root using a standard suction electrode. Single-fiber action potentials were amplified and displayed on a digital storage oscilloscope and recorded for computer analysis (fig. 1). Stimulus-evoked activity in individual axons was monitored, and the latency between stimulus and action potential was measured. Conduction velocity (CV) was calculated from measurements of conduction latency and length of axon between stimulating and recording electrodes (CV [m/s]= conduction distance [mm] divided by conduction latency [ms]). Control CV measurements from each axon were recorded for 20–30 min to ensure stability of the preparation before exposure to ICG (Cardio-Green®).

In a preliminary study, single ICG doses of 300 μg in 0.1 ml were injected into the constantly flowing aCSF as it entered the perfusion chamber. Each injection was repeated at 3-min intervals for a total of three injections. Based on the aCSF flow rate and injection time, we calculated that each bolus dose would transiently produce an ICG concentration ∼2 times higher than the blood concentration following the standard recommended dose and ∼100 times higher than the minimum transcutaneously detectable dose. This technique was designed to mimic the effects of dilution and binding believed to occur after subarachnoid injection of ICG in humans. Stimulus threshold, conduction latency, and ectopic spike generation were measured at 1-min intervals for 30 min.

In the subsequent series of experiments, isolated but otherwise intact DRs were continuously perfused with aCSF containing ICG at one of three different concentrations: group 1, 28.6 μm (n = 10); group 2, 143 μm (n = 8); and group 3, 286 μm (n = 8). These concentrations represented the 1-fold, 5-fold, and 10-fold CSF concentrations that we estimated would be present in the lumbar CSF after the inadvertent intrathecal injection of a minimum ICG test dose. This estimate was based on the following assumptions: (1) the minimum detectable intravenous ICG dose is 7 μg/kg (0.5 mg/70 kg);17(2) a typical test dose should contain 1 mg ICG to provide a margin of detection safety; and (3) an intrathecal injection of 1 mg ICG could equilibrate in ∼50 ml of spinal CSF producing a concentration of 26 μm. However, with minimal mixing the equilibration volume could be significantly less, resulting in the exposure of spinal roots to higher ICG concentrations.

Following a series of control measurements, axons in the intact DR were exposed to the ICG containing aCSF for 25 min while repeated measurements of threshold, conduction latency, and the frequency of ectopic impulses were made. A washout period followed for about 60 min during which all measurements were continued. Only one series of measurements was obtained from one axon in each nerve root, and each axon served as its own control. Axons with unstable control measurements were not tested. As expected, the exposed portion of each intact nerve root was stained green by the ICG, and this coloration persisted during the washout period.

In a separate control study to examine the possible contribution of sodium iodide to the observed effects, single DR axons (four myelinated and two unmyelinated) were exposed to sodium iodide in aCSF at a concentration of 69 μm. This represents the approximate concentration that would be encountered with our highest ICG dose. Interval measurements were made during 25 min of sodium iodide exposure and for a 30-min washout period. Again, only one series of measurements was obtained from one axon in each nerve root.

For data analysis, DR axons were divided into two groups based on their CV characteristics. Those DR axons with a CV greater than 3 m/s were considered to be myelinated (A-fiber), whereas those with a CV less than 1.5 m/s were presumed to be unmyelinated (C-fiber). Axons with intermediate CV were not studied. The Fisher exact text was used to measure the statistical significance of the observed differences between treatment groups. Given the small number of axons studied and the apparent absence of a significant dose-response effect, we grouped the ICG-exposed axons together for final comparison with the control group.

In the preliminary experiments, exposure of isolated DRs to a series of three 300-μg bolus doses of ICG produced conduction block in two of four myelinated and two of two unmyelinated DR axons without recovery during the subsequent 30-min washout period. Spontaneous bursting activity was observed in the other two myelinated DR axons after ICG exposure (fig. 2). During the 30-min washout period this bursting activity stopped but could be started again with reexposure to ICG.

In the next series of experiments, 26 DR axons in 26 DRs from 14 rats were continuously exposed for 25 min to one of three concentrations of ICG in aCSF. The length of axon (root) exposed to ICG ranged from 17 to 26 mm. The CV for the myelinated axons ranged from 4.09 to 22.81 m/s and for the unmyelinated axons from 0.66 to 0.90 m/s. Twenty-one of these axons (81%) were affected by exposure to ICG. Conduction block occurred in 14 of 23 myelinated axons (significantly different from control), and no axon recovered from this conduction block during the 60-min postexposure observation period. The onset of block in A-fibers ranged from 8 to 50 min, averaging 23 min. In some axons, conduction block did not occur until the postexposure washout period. None of the three unmyelinated axons tested were blocked by ICG. In 5 of 23 myelinated axons and in 2 of 3 unmyelinated axons, exposure to ICG produced spontaneous bursts of action potentials. This spontaneous high frequency activity always began after ICG washout and continued throughout the postexposure period or until conduction block occurred. The incidence of conduction block and bursting activity (fig. 3) was not significantly different among the three groups but was significantly different when the grouped incidence was compared with the control. Three of 23 myelinated axons and 1 of 3 unmyelinated axons were minimally affected by ICG exposure, demonstrating only decreased CV to a mean of 62% (range, 42–77%) and 44% of control CV, respectively. CV slowing was produced by ICG in all 26 axons in an apparently dose-dependent manner (fig. 4). In many cases at the conclusion of the washout period (60 min), other axons in the intact portion of that DR were isolated and briefly studied. That was done to examine the possibility that the nerve dissection itself had somehow made the isolated axons uniquely vulnerable to ICG toxicity. However, in every DR screened in this manner, only blocked axons or axons exhibiting spontaneous bursting activity were encountered.

In the sodium iodide control study, four myelinated and two unmyelinated DR axons were exposed to a high concentration of sodium iodide (69 μm). No effects on action potential generation or CV were observed during the 25-min exposure period or during the subsequent 30-min washout period.

The results of this study demonstrate that intrathecal ICG may be potentially neurotoxic in concentrations likely to be encountered when the dye is used as a marker for inadvertent intravascular injection during epidural anesthesia. Many DR axons briefly exposed to ICG developed either persistent conduction block or spontaneous bursting activity. These effects could not be attributed to the iodide present in the commercial ICG preparation because exposure to iodide in control experiments was without effect. Clinically, intrathecal ICG could be expected to produce a combination of potentially painful paresthesias and persistent sensory loss. If motor axons in ventral roots are similarly affected, then intrathecal ICG could produce paralysis.

In a recent review of epidural injection safety procedures, the authors state that “the most significant hazard of epidural blockade is unrecognized unintentional intravascular injection,”5which occurs with an estimated frequency of 2% in the general population and 7–8.5% in the obstetric population. Because the consequences of bolus intravascular injection of local anesthetic can be disastrous, there is a clear need for a sensitive and specific test to detect inadvertent intravascular needle or catheter placement. Currently no such test exists. Mulroy et al.  5recommend using a combination of procedures, including aspiration, incremental injections, and use of epinephrine-containing test doses in nonlaboring and other appropriately selected patients, while minimizing the concentration and dose of the local anesthetic.

The search for the perfect marker to objectively and safely detect intravascular misplacement of an epidural needle or catheter is ongoing. The dye ICG has been suggested as an alternative to commonly used markers including epinephrine. 17,18Loehlein and Schmidt 17were able to detect intravenously administered ICG at concentrations as low as 7 μg/kg by transcutaneous photometry at 808 nm, whereas Laurito and Chen 18used a spectrophotometric technique at 770 nm to detect the intravenous injection of ICG at a dose of 500 μg/kg. Both groups proposed mixing ICG with local anesthetics to provide a rapid and continuous means of detecting accidental intravascular injections.

Indocyanine green is a water-soluble dye with a molecular weight of 775. After intravenous administration it is rapidly bound to plasma proteins and cannot be detected in CSF. 19It is confined to the vascular compartment, is not metabolized, and is excreted only by the liver. Given these attributes, it has been used extensively to assess liver function, cardiac output, cerebral blood flow, 23blood volume, 24,25and as a contrast agent in imaging studies. ICG has an impressive safety record with relatively few adverse reactions reported since it was first approved for use in humans in 1956. 19,26Most reactions consist of urticaria, sensations of warmth, headache, or nausea. More rarely, patients have become hypotensive or dyspneic. For example, in a series of 1,923 ICG injections, only 8 patients experienced an adverse reaction. 19Two patients developed urticaria and were treated with oral diphenhydramine. Two others developed vasovagal-like reactions from which they recovered spontaneously. One patient became severely hypotensive and was admitted for overnight observation. The remaining three patients had mild reactions consisting of nausea or injection-related discomfort. In a review of the world literature, only eight deaths have been attributed to ICG administration, and they all appear to be the result of anaphylactic reactions. 19 

The proposal to use ICG as part of an epidural test dose to detect intravascular injection is appealing. It has many of the attributes of an ideal marker. It would be simple to incorporate into standard test doses of local anesthetic; it can be safely injected intravascularly; and its appearance in the circulation can be simply detected using a noninvasive photometric technique. However, the results of the present study have revealed a potentially neurotoxic effect of ICG on DR axons, which could preclude its use in this application.

The authors thank Alfred Schmidt, M.D., for his assistance in the experiments. Dr. Schmidt was a visiting anesthesiologist, Department of Anesthesia, Technical University of Aachen, Aachen, Germany. Indocyanine green was generously supplied by Becton Dickinson Microbiology Systems, Cockeysville, Maryland.