We appreciate the comments of Drs. Lambert and Strichartz. Their letter raises a number of valid points regarding the utility and value of in vitro  experimentation. However, they seem to confuse utility with clinical relevance, trivializing the distinction between isolated fragments of nerve and more complex physiologic systems, and underestimating the disparity between absolute concentrations determined in vitro  and relevant concentrations in an intact animal.

The purpose of our study was to determine the relative neurotoxicity of intrathecally administered chloroprocaine and bisulfite. Although other investigators had previously evaluated these compounds, their comparative toxicity had never been established. Our results demonstrated that chloroprocaine was potentially neurotoxic when administered at a clinically relevant concentration, whereas bisulfite seemed to be neuroprotective, a rather surprising finding given the view held by many that bisulfite, not chloroprocaine, was responsible for the early clinical injuries associated with Nesacaine-CE. This prevailing view was based on a number of studies but principally the work of Gissen et al.  1Using an isolated nerve model, they reported that exposure to chloroprocaine with bisulfite at a pH of 3 produced irreversible conduction failure, whereas the same solution at a pH of 7.3 resulted in recovery; irreversible block also occurred with exposure to bisulfite without chloroprocaine, but only at a low pH. Our discussion of possible sources for these conflicting data included factors unique to the model of Gissen et al.  that present limitations when extrapolating from these in vitro  studies to intact mammalian systems.

It is puzzling that Drs. Lambert and Strichartz take our discussion of the limitations of the experiments of Gissen et al.  (with which they completely agree) to represent a global condemnation of in vitro  experimentation. We do not—nor would any reasonable person—dispute the critical role of in vitro  experimentation in scientific inquiry and drug development. Indeed, that muscle relaxants were first tested in vitro  grossly understates the utility of such experiments—it would be fair to say that without in vitro  studies, few drugs would exist. We have personally used a variety of in vitro  models in our explorations of anesthetic neurotoxicity. In addition to studies of conduction failure in isolated nerve, we have investigated the role of intracellular calcium using dorsal root ganglia cell culture,2an in vitro  system perhaps more remotely linked to clinical practice. Moreover, included in the reference list of Drs. Lambert and Strichartz is an article containing studies we performed in a plastic tube simulating the subarachnoid space.3Nonetheless, although in vitro  experimentation requires no defense, the limitations imposed by the unique characteristics of these models must be considered.

Drs. Lambert and Strichartz defend the use of conduction failure by Gissen et al.  as a surrogate outcome based on the fact that clinical deficits likely arise from irreversible conduction loss. We agree that this is a logical physiologic endpoint. However, although clinical deficits may arise from conduction failure, the corollary is not necessarily true, at least not for studies conducted on nerve fragments—there are many things capable of producing conduction loss in this unstable in vitro  system that would not impact an intact animal. Extrapolation must therefore be made cautiously, a point apparently recognized by Drs. Lambert and Strichartz as they have previously commented: “It is possible that the acute irreversible loss of conduction occurs by different mechanisms than those yielding slower developing prolonged conduction deficits.”4 

They state that lack of knowledge regarding the relevant concentration in vitro “does not preclude” investigation. We agree. However, it does place constraints on the data that again must be considered when extrapolating to intact physiologic systems. This point can be readily appreciated by examining the anesthetic concentrations required to produce conduction block in their isolated sciatic nerve preparation.4In this in vitro  model, tetracaine is 100 times more potent than lidocaine and 18 times more potent than bupivcaine.4In contrast, despite marked differences in methodology, potency ratios determined in vivo  closely parallel clinical practice.5,6 

Drs. Lambert and Strichartz question the use of infusion in our model, asking what clinical (and physiologic) scenario this represents. This model was developed to investigate anesthetic neurotoxicity after a series of reports of clinical injury associated with continuous spinal anesthesia.7,8Substantial clinical7and experimental3evidence suggested that maldistribution was an important etiologic factor in injury, i.e. , maldistribution resulted in high anesthetic concentrations within a restricted area of the subarachnoid space, unmasking the potential toxicity of the anesthetic agents. Accordingly, to investigate factors that may affect such injuries, we developed a model in which a restricted sacral distribution is deliberately produced, in part, by administering drug by infusion.9,10In its exposure of neural elements to very high, albeit regional, concentrations of anesthetic solutions, our model parallels the clinical injuries that occurred with Nesacaine-CE. Maldistribution provides other advantages as well. The limited spread avoids hemodynamic changes, and unpublished data from our laboratory demonstrate that blood pressure is virtually unaffected with this method of administration. In addition, because deficits are limited caudally, animals require only minimal care, and they can be maintained for prolonged periods after injury without generating concern for their welfare. This in turn permits extensive functional and histologic studies. Nonetheless, the point raised by Drs. Lambert and Strichartz is valid—the extent to which our model differs from the early cases in which large bolus doses of Nesacaine-CE were administered intrathecally is a limitation that must be considered, at least with respect to these particular clinical injuries.

Dr. Baker’s letter raises some critical points that reinforce a seventh limitation of the data of Gissen et al.  that were not included in the comments of Drs. Lambert and Strichartz and that apply to our data as well. It was the postulate of Gissen et al.  that bisulfite induced injury through liberation of sulfur dioxide, a reaction favored by low pH.1However, as Gissen et al.  note, the normal body economy is well protected by enzymatic conversion of sulfites to less toxic sulfates. But it is possible that enzyme activity was severely depressed in these in vitro  experiments, given that they were conducted at room temperature on segments of disrupted nerve exposed to a pH equivalent to that of the test solution. With respect to the latter point, despite the comments of Drs. Lambert and Strichartz regarding subarachnoid distribution, it is highly unlikely that even large volumes of solution administered intrathecally would be completely undiluted or that pH would be completely unaffected by cerebral spinal fluid and cellular buffers. Therefore, there are a number of factors that may make the model of Gissen et al.  uniquely vulnerable to bisulfite toxicity. In contrast, our model may be relatively insensitive because of higher concentrations of sulfite oxidase in rats compared with humans.11Clearly, much more information is needed to adequately appreciate the clinical toxicity and the neuroprotective potential of such compounds.

In short, the strengths, limitations, and unique characteristics of the various experimental models must be considered when interpreting the data they generate, and our comments regarding the data of Gissen et al.  served to highlight the unique aspects of their model that could potentially generate results at variance to clinical reality. These comments can be summed up well by a passage from Drs. Lambert and Strichartz’ published investigation of anesthetic toxicity in isolated nerve: “However, caution should be used in quantitatively extending these toxic effects to mammalian nerves in vivo .”4 

* University of California, San Francisco, California. kdrasner@anesthesia.ucsf.edu

Gissen A, Datta S, Lambert D: The chloroprocaine controversy: II. Is chloroprocaine neurotoxic? Regional Anesthesia 1984; 9:135–44
Gold MS, Reichling DB, Hampl KF, Drasner K, Levine JD: Lidocaine toxicity in primary afferent neurons from the rat. J Pharmacol Exp Ther 1998; 285:413–21
Rigler ML, Drasner K: Distribution of catheter-injected local anesthetic in a model of the subarachnoid space. Anesthesiology 1991; 75:684–92
Lambert LA, Lambert DH, Strichartz GR: Irreversible conduction block in isolated nerve by high concentrations of local anesthetics. Anesthesiology 1994; 80:1082–93
Sakura S, Bollen AW, Ciriales R, Drasner K: Local anesthetic neurotoxicity does not result from blockade of voltage-gated sodium channels. Anesth Analg 1995; 81:338–46
Ready LB, Plumer MH, Haschke RH, Austin E, Sumi SM: Neurotoxicity of intrathecal local anesthetics in rabbits. Anesthesiology 1985; 63:364–70
Rigler ML, Drasner K, Krejcie TC, Yelich SJ, Scholnick FT, DeFontes J, Bohner D: Cauda equina syndrome after continuous spinal anesthesia. Anesth Analg 1991; 72:275–81
FDA Safety Alert: Cauda equina syndrome associated with the use of small-bore catheters in continuous spinal anesthesia. Rockville, MD, Food and Drug Administration, May 29, 1992
Drasner K, Sakura S, Chan VW, Bollen AW, Ciriales R: Persistent sacral sensory deficit induced by intrathecal local anesthetic infusion in the rat. Anesthesiology 1994; 80:847–52
Sakura S, Hashimoto K, Bollen AW, Ciriales R, Drasner K: Intrathecal catheterization in the rat: Improved technique for morphologic analysis of drug-induced injury. Anesthesiology 1996; 85:1184–9
Gunnison AF, Bresnahan CA, Palmes ED: Comparative sulfite metabolism in the rat, rabbit and rhesus monkey. Toxicol Appl Pharmacol 1977; 42:99–109