JOURNAL OF NEUROTRAUMA Volume 9, Number 2, 1992 Mary Ann Liebert, Inc., Publishers
Rapid Quantification of Tissue Damage for Assessing Acute Spinal Cord Injury Therapy WISE YOUNG
of the National Acute Spinal Cord Injury Study (NASCIS II) have studies of acute spinal cord injury treatments in several ways. First, NASCIS II showed clearly the importance of detailed knowledge of drug doses for clinical trials. In fact, the previous trial failed to show treatment effects, at least in part, because insufficient dose-response data were available from animal studies. Systematic dose-response tests are essential. Second, because an effective treatment has been found, e.g., methylprednisolone (MP), all spinal cord injury studies must now compare treatments not only against placebo but against MP as well. Third, because it will be difficult to convince both patients and clinical investigators to use an untested drug in the place of MP, it is likely that future clinical trials will involve the addition of a drug in combination with or after high-dose MP. Thus, preclinical testing must also examine drug combinations. Fourth, NASCIS II showed that the timing of treatment is crucial to beneficial effects. Patients treated more than 8 h after injury not only showed no beneficial effect but actually may have gotten worse as a result. One consequence of these developments is that laboratory investigators must bear the burden of not only showing that a treatment is effective but also of finding the optimal dose, the best dose combination, the treatment period, and the relationship of the treatment period with injury severity before that treatment will be considered for clinical trial. This burden is onerous. Even examination of a minimum set of variables—three doses, three treatment times, three treatment durations, three injury severities, three combinations with MP, and vehicle controls—would require the testing of about 100 treatment groups. Many currently used spinal cord injury models and outcome measures require 10-20 animals per treatment group to detect behavioral or biochemical outcome differences. Even if we assume that the model is sufficiently reproducible to detect 10% outcome differences with only 10 animals, a comprehensive study would require over 1000 experiments. Most laboratories cannot carry out and evaluate more than one or two spinal cord injuries per day. No single laboratory can do the tour de force study necessary to collect all the data required to take a treatment to clinical trial. This problem is a very serious one and could arrest the renaissance of acute spinal cord injury research that has taken place in the past decade. Several approaches to this problem should be considered. First, multicenter animal studies can be carried out. By spreading the burden of animal testing over several different laboratory groups, it may be possible to establish the best dose, timing of treatment, combination therapies, and injury severity parameters within a shorter period of time. Second, a model must be developed that allows precise monitoring of injury conditions so that smaller treatment groups can be used and outcomes closely correlated with injury severity. Third, treatment testing can be carried out in stages, with an initial screening step to identify the injury and treatment parameters that are most likely to yield positive results. More detailed anatomic, behavioral, and neurophysiologic outcome testing can then be focused on the most promising therapies.
Thecomplicated
recently published results
Department of Neurosurgery, New York University Medical Center, New York, New York. 151
YOUNG The New York University Neurosurgery Laboratory has devoted considerable time and effort to developing experimental spinal cord injury model, a method that allows rapid screening and testing in a large number of animals. The model of graded spinal cord contusion in the rat has proven to be very reproducible and efficient. A simple and inexpensive apparatus precisely monitors all the relevant parameters of the injury. A method based on measuring total tissue sodium and potassium has been developed for quantifying tissue damage in the spinal cord with a standard deviation of less than 5%. Under normal conditions, there are high concentrations of intracellular potassium and extracellular sodium, but injury essentially causes cells to join the extracellular space. Thus, total tissue sodium concentrations ([Na]t) will rise while tissue potassium concentrations ([K]t) will fall. Previous studies in our laboratory and others have shown that these ionic concentrations correlate with injury severity. However, the theoretical relationship between the ionic shifts and tissue damage was not well understood until recently. This relationship is based on a single assumption, that sodium and potassium together constitute more than 95% of osmolarity of tissue fluids. Since 1 mJW of [Na] [K] difference across membranes can generate nearly 20 mm Hg of pressure, it is reasonable to assume that the sum of Na and K concentrations is the same inside and outside of cells. an
=
[Na]i This
+
[K]i
=
[Na]e
+
[K]e.
implies that transmembrane Na and K gradients must be equal. [Na]i -
From definition,
we
[Na]e
=
[K]i
[K]e
G
=
-
also know that
[Na]t
x
Vt
=
[Na]i
x
Vi +
[Naje
[K]i
x
Vi +
[K]e
Ve
x
and
[K]t where Vt, two
x
Vt
=
Ve
x
Vi, and Ve are, respectively, the total, intracellular, and extracellular volumes. Subtracting these
equations and substituting G gives [Na]t
-
[K]t
=
[Na]e
-
[K]e
+
G
x
Vi/Vt
This equation essentially states that the difference of total tissue Na and K concentrations is linearly related to the ratio of cellular and tissue volume with a slope of G and a y-intercept of [Na]e [K]e. We have tested this relationship extensively and shown that [Na]t [K]t accurately and linearly predicts the volume of intact cells in a large variety of tissues, including brain and spinal cord. For example, [Na]t [K]t correlated with blood hematocrit and with lesion volume in injured spinal cords with a correlation coefficient over 0.96. Using this approach, we have quantified the tissue damage in rat spinal cords at 6 and 24 h after graded spinal contusions. Based on data from more than 1000 experiments, we have been able to demonstrate that the measurements are extremely accurate and produce less than 5% standard coefficients of variance in repeated measures. We can easily, for example, distinguish injuries caused by a 10 g weight dropped 12.5, 25, and 50 cm with fewer than five animals per group. Using this method, we have identified the critical physical parameters of contusion for prediction of tissue damage. The model and [Na]t [K]t method was then used to evaluate three drug therapies of acute spinal cord monosialic tirilazad (Tz), ganglioside (GM1), and cyclosporin (Cy). These drugs were tested alone injury: and in combination with MP on tissue damage at 24 h after graded injury in approximately 1000 rats with systematic dose-response and combination therapy testing. The optimal time window for therapy has not yet been studied. Tz is a synthetic 21-aminosteroid with no corticosteroid activity, which is about 100 times more potent than MP as a lipid peroxidation inhibitor. GM1 was reported recently to improve neurologic recovery in patients after spinal cord injury. Tz modestly reduced the lesion volume but only in rats with mild to —
—
—
—
152
END POINT MEASURES: BIOCHEMICAL moderate spinal contusions. MP significantly reduced tissue damage in mild, moderate, and severe injuries. MP followed by Tz was more effective than Tz alone and marginally more effective than MP alone. However, since Tz does not have the steroid side effects of MP, the MP plus Tz treatment is attractive and will be tested in a clinical trial. GM1 alone had no significant effect on tissue damage at any of the doses examined. GM1 given concomitantly with MP, however, eliminated the beneficial effects of MP. The best treatment evaluated in this study was Cy plus Tz, which reduced tissue damage more consistently at all injury severities than did MP alone. Our experience shows that the method of graded weight drop contusion with monitoring of the impact will generate very reproducible tissue damage in rats. In addition, at least three treatments effectively reduce the lesion size at 24 h. Comparison to the tissue damage over time argues strongly for the presence of progressive increases in the lesion size over a 24 h period. The finding that MP plus Tz effectively reduces tissue damage is very encouraging. The observation that Cy is beneficial but is most effective when combined with Tz is exciting and suggest for the first time that Cy binds immune receptors. Finally, the findings for Cy plus Tz present the first drug combination that is an improvement over MP alone. Additional experiments are needed to resolve the mechanisms of action of these drugs.
REFERENCES BRACKEN, M.B., SHEPARD, M.J., COLLINS, W.F., HOLFORD, T.R., YOUNG, W., BASKIN, D.S., EISENBERG, H.M., FLAMM, E., LEO-SUMMERS, L., MAROON, J., MARSHALL, L.F., PEROT, P.L., PIEPMEIER, J., SONNTAG, V.K.H., WAGNER, F.C., WILBERGER, J.E., and WINN, H.R. (1990). A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute
spinal cord injury. N. Engl. J. Med. 322,
1405-1411.
BRACKEN, M.B., SHEPARD, M.J., COLLINS, W.F., HOLFORD, T.R., BASKIN, D.S., EISENBERG, H.M., FLAMM, E., LEO-SUMMERS, L., MAROON, J., MARSHALL, L.F., PEROT, PL., PIEPMEIER, J., SONNTAG, V.K.H., WAGNER, F.C., WILBERGER, J.E., WINN, H.R., and YOUNG, W. (1992). Methylprednisolone naloxone in the treatment of acute spinal cord injury: One year follow-up data. Results of the second National Acute Spinal Cord Injury Study. J. Neurosurg. 76, 23-31. GEISLER, F.H., DORSEY, F.C., and COLEMAN, W.P. ( 1991 ). Recovery of motor function after spinal cord injury: A randomized, placebo-controlled trial with GM-1 ganglioside. N. Engl. J. Med. 324, 1829-1838. or
Address reprint requests to: Wise Young, M.D., Ph.D.
Department of Neurosurgery University Medical School
New York
550 First Avenue New York, NY 10016
DISCUSSION Dr. Wrathall: What is the influence of the red blood cells that are trapped? They replace neural tissue, and I presume are counted as "intact tissue." Dr. Young: Infiltration of RBCs would contribute to the sodium minus potassium difference so that there would appear to be intact tissue. This measurement allows you to distinguish only the cellular volume fraction. This is very straightforward. The RBCs also will never reach beyond the hematocrit of blood, which in a rat is typically about 35%. I have never seen the cell volume fraction at the impact site go below 30%. Maybe it is because of blood entering the tissue. Dr. Wrathall: I must admit that we have been disappointed in the correlation between behavioral or electrophysiologic measurements and tissue damage, and we have looked at many spinal cords. But I do not trust our histologie measurement of the tissue sufficiently so that I can show a graph relating sodium minus potassium and histology. This is a destructive technique, and once we kill the animal for this measurement, we 153
YOUNG cannot then look at the
same
spinal cord 3
weeks later for
histologie changes.
It is my
general impression,
however, that it predicts the degree of histologically apparent damage at 2 or 3 weeks after injury and certainly
predicts our 3 or 4 week injury severity levels.
Dr. Tator: The increase in total sodium at the injury site is somehow balanced by a major increase in calcium intracellularly. If one were to examine only sodium minus potassium as the index of injury and treatment, are we going to miss something? Can we see effects for drugs that act primarily at the level of calcium? Dr. Young: Of course. If a drug saves cells, it would show up in the sodium minus potassium difference. Na minus K directly reflects the volume of cells that can maintain their sodium or potassium gradients across a membrane. By the way, the amount of calcium is very small compared to sodium and potassium, about 1 or 2mM. Dr. Tator: That's why I am concerned about looking at only sodium and potassium. If the drug acts on calcium and you don't need huge amounts of calcium inside vs outside to kill cells, then what? Dr. Young: What we are essentially measuring is the hematocrit of the tissue. We are using sodium minus potassium as a measurement of the cell volume fraction, and we have derived this equation from one simple assumption. As long as we assume isotonicity (Na K) is proportional to Vi/Vt and Vi/Vt is the cell volume fraction of the tissue, we are saying that if you hit the spinal cord, there will be loss of cells. The cells will join the extracellular space, and if you have any treatment that saves those cells, you have less change in (Na K). It does not address mechanisms. It is a simple measure of cells that are intact in the tissue. Dr. Stokes: We are dealing with an extremely heterogeneous tissue, and the cells we are interested in may not be the ones that are showing these dramatic shifts. Intuitively, getting an index of damage is important, but I think it must be closely tied to relative histopathologic index in specific cell types. Dr. Young: No. You certainly do not do that in your laboratory when you look at necrosis. I am telling you that if you are looking for necrosis in the spinal cord and it takes you 6 weeks to do 10 animals to map out the necrotic area, you can do that in 1 day using the sodium minus potassium method. Dr. Stokes: You don't agree that this index would have to be correlated with histopathology, axonal types, or neurons? Dr. Young: That's why we do electrophysiology. This method is a simple, straightforward, robust measurement of cell volume fraction. It is equivalent to doing serial sections of the tissue, mapping out the necrotic area straight through, and then summing it up. I presented a slide showing a direct linear correlation between the necrotic area in the tissue and the sodium minus potassium changes. In fact, it predicts precisely both the extracellular ionic differences and the expected gradient across the membranes. Dr. Hsu: We have been looking into vasogenic edema in the brain and spinal cord. Evolution of edema in the spinal cord after trauma is similar to your sodium minus potassium results. Maybe you have developed another way to determine vasogenic edema. Dr. Young: We have measured albumin entry into the tissue and correlated with sodium minus potassium, and indeed there is a nice correlation. I think that necrosis clearly correlates with the degree of endothelial breakdown in the tissue. But the nice thing about sodium minus potassium is that this is a relationship that predicts precisely the volume of intact cells. Dr. Faden: The highest predictor in terms of correlation coefficient (around 0.96) with regard to outcome and injury severity is tissue magnesium. Tissue magnesium also correlates highly to lipid hydrolysis, so there is both a mechanism and a predictor that seem to be very sensitive to drug treatment effects and to degree of injury. I wonder if you could comment on the magnesium component and whether you think there might be an avenue for using both the sodium minus potassium and magnesium together as a combined predictor. Dr. Young: We measure magnesium in all our tissue specimens. It does correlate extremely well with the sodium minus potassium. I believe magnesium reflects the amount of lipid loss in the tissue, and lipid loss is one of the primary markers of necrosis. By using extracellular ion selective microelectrodes in the rat spinal cord injury model, we have found that the rat clears out the sodium and potassium very rapidly from extracellular space. Extracellular potassium is normalized by 20 min after an impact injury, even in severe injuries. I believe that you can assume that sodium minus potassium in the extracellular space is equal to plasma levels by 30 min after injury, and from that point, you can simply analyze the data. I must put in a caveat, and that is we are making the cellular volume fraction calculations based on the value of the ion —
—
154
END POINT MEASURES: BIOCHEMICAL
gradient across membranes (G), and there is no easy way of determining this. As it is turning out, the ionic gradient probably is low for as long as 6 h after injury. We always take a segment of cord that's far away from the injury site to estimate a G value, and we also take specimens from other parts of the body to be sure that we do not have an animal with a G value that is remarkably different. We are still working on finding a method where we can use the G value from a normal piece of tissue to normalize our data at the injury site so we can apply earlier analyses.
155
Rapid Quantification of Tissue Damage for Assessing Acute Spinal Cord Injury Therapy WISE YOUNG
of the National Acute Spinal Cord Injury Study (NASCIS II) have studies of acute spinal cord injury treatments in several ways. First, NASCIS II showed clearly the importance of detailed knowledge of drug doses for clinical trials. In fact, the previous trial failed to show treatment effects, at least in part, because insufficient dose-response data were available from animal studies. Systematic dose-response tests are essential. Second, because an effective treatment has been found, e.g., methylprednisolone (MP), all spinal cord injury studies must now compare treatments not only against placebo but against MP as well. Third, because it will be difficult to convince both patients and clinical investigators to use an untested drug in the place of MP, it is likely that future clinical trials will involve the addition of a drug in combination with or after high-dose MP. Thus, preclinical testing must also examine drug combinations. Fourth, NASCIS II showed that the timing of treatment is crucial to beneficial effects. Patients treated more than 8 h after injury not only showed no beneficial effect but actually may have gotten worse as a result. One consequence of these developments is that laboratory investigators must bear the burden of not only showing that a treatment is effective but also of finding the optimal dose, the best dose combination, the treatment period, and the relationship of the treatment period with injury severity before that treatment will be considered for clinical trial. This burden is onerous. Even examination of a minimum set of variables—three doses, three treatment times, three treatment durations, three injury severities, three combinations with MP, and vehicle controls—would require the testing of about 100 treatment groups. Many currently used spinal cord injury models and outcome measures require 10-20 animals per treatment group to detect behavioral or biochemical outcome differences. Even if we assume that the model is sufficiently reproducible to detect 10% outcome differences with only 10 animals, a comprehensive study would require over 1000 experiments. Most laboratories cannot carry out and evaluate more than one or two spinal cord injuries per day. No single laboratory can do the tour de force study necessary to collect all the data required to take a treatment to clinical trial. This problem is a very serious one and could arrest the renaissance of acute spinal cord injury research that has taken place in the past decade. Several approaches to this problem should be considered. First, multicenter animal studies can be carried out. By spreading the burden of animal testing over several different laboratory groups, it may be possible to establish the best dose, timing of treatment, combination therapies, and injury severity parameters within a shorter period of time. Second, a model must be developed that allows precise monitoring of injury conditions so that smaller treatment groups can be used and outcomes closely correlated with injury severity. Third, treatment testing can be carried out in stages, with an initial screening step to identify the injury and treatment parameters that are most likely to yield positive results. More detailed anatomic, behavioral, and neurophysiologic outcome testing can then be focused on the most promising therapies.
Thecomplicated
recently published results
Department of Neurosurgery, New York University Medical Center, New York, New York. 151
YOUNG The New York University Neurosurgery Laboratory has devoted considerable time and effort to developing experimental spinal cord injury model, a method that allows rapid screening and testing in a large number of animals. The model of graded spinal cord contusion in the rat has proven to be very reproducible and efficient. A simple and inexpensive apparatus precisely monitors all the relevant parameters of the injury. A method based on measuring total tissue sodium and potassium has been developed for quantifying tissue damage in the spinal cord with a standard deviation of less than 5%. Under normal conditions, there are high concentrations of intracellular potassium and extracellular sodium, but injury essentially causes cells to join the extracellular space. Thus, total tissue sodium concentrations ([Na]t) will rise while tissue potassium concentrations ([K]t) will fall. Previous studies in our laboratory and others have shown that these ionic concentrations correlate with injury severity. However, the theoretical relationship between the ionic shifts and tissue damage was not well understood until recently. This relationship is based on a single assumption, that sodium and potassium together constitute more than 95% of osmolarity of tissue fluids. Since 1 mJW of [Na] [K] difference across membranes can generate nearly 20 mm Hg of pressure, it is reasonable to assume that the sum of Na and K concentrations is the same inside and outside of cells. an
=
[Na]i This
+
[K]i
=
[Na]e
+
[K]e.
implies that transmembrane Na and K gradients must be equal. [Na]i -
From definition,
we
[Na]e
=
[K]i
[K]e
G
=
-
also know that
[Na]t
x
Vt
=
[Na]i
x
Vi +
[Naje
[K]i
x
Vi +
[K]e
Ve
x
and
[K]t where Vt, two
x
Vt
=
Ve
x
Vi, and Ve are, respectively, the total, intracellular, and extracellular volumes. Subtracting these
equations and substituting G gives [Na]t
-
[K]t
=
[Na]e
-
[K]e
+
G
x
Vi/Vt
This equation essentially states that the difference of total tissue Na and K concentrations is linearly related to the ratio of cellular and tissue volume with a slope of G and a y-intercept of [Na]e [K]e. We have tested this relationship extensively and shown that [Na]t [K]t accurately and linearly predicts the volume of intact cells in a large variety of tissues, including brain and spinal cord. For example, [Na]t [K]t correlated with blood hematocrit and with lesion volume in injured spinal cords with a correlation coefficient over 0.96. Using this approach, we have quantified the tissue damage in rat spinal cords at 6 and 24 h after graded spinal contusions. Based on data from more than 1000 experiments, we have been able to demonstrate that the measurements are extremely accurate and produce less than 5% standard coefficients of variance in repeated measures. We can easily, for example, distinguish injuries caused by a 10 g weight dropped 12.5, 25, and 50 cm with fewer than five animals per group. Using this method, we have identified the critical physical parameters of contusion for prediction of tissue damage. The model and [Na]t [K]t method was then used to evaluate three drug therapies of acute spinal cord monosialic tirilazad (Tz), ganglioside (GM1), and cyclosporin (Cy). These drugs were tested alone injury: and in combination with MP on tissue damage at 24 h after graded injury in approximately 1000 rats with systematic dose-response and combination therapy testing. The optimal time window for therapy has not yet been studied. Tz is a synthetic 21-aminosteroid with no corticosteroid activity, which is about 100 times more potent than MP as a lipid peroxidation inhibitor. GM1 was reported recently to improve neurologic recovery in patients after spinal cord injury. Tz modestly reduced the lesion volume but only in rats with mild to —
—
—
—
152
END POINT MEASURES: BIOCHEMICAL moderate spinal contusions. MP significantly reduced tissue damage in mild, moderate, and severe injuries. MP followed by Tz was more effective than Tz alone and marginally more effective than MP alone. However, since Tz does not have the steroid side effects of MP, the MP plus Tz treatment is attractive and will be tested in a clinical trial. GM1 alone had no significant effect on tissue damage at any of the doses examined. GM1 given concomitantly with MP, however, eliminated the beneficial effects of MP. The best treatment evaluated in this study was Cy plus Tz, which reduced tissue damage more consistently at all injury severities than did MP alone. Our experience shows that the method of graded weight drop contusion with monitoring of the impact will generate very reproducible tissue damage in rats. In addition, at least three treatments effectively reduce the lesion size at 24 h. Comparison to the tissue damage over time argues strongly for the presence of progressive increases in the lesion size over a 24 h period. The finding that MP plus Tz effectively reduces tissue damage is very encouraging. The observation that Cy is beneficial but is most effective when combined with Tz is exciting and suggest for the first time that Cy binds immune receptors. Finally, the findings for Cy plus Tz present the first drug combination that is an improvement over MP alone. Additional experiments are needed to resolve the mechanisms of action of these drugs.
REFERENCES BRACKEN, M.B., SHEPARD, M.J., COLLINS, W.F., HOLFORD, T.R., YOUNG, W., BASKIN, D.S., EISENBERG, H.M., FLAMM, E., LEO-SUMMERS, L., MAROON, J., MARSHALL, L.F., PEROT, P.L., PIEPMEIER, J., SONNTAG, V.K.H., WAGNER, F.C., WILBERGER, J.E., and WINN, H.R. (1990). A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute
spinal cord injury. N. Engl. J. Med. 322,
1405-1411.
BRACKEN, M.B., SHEPARD, M.J., COLLINS, W.F., HOLFORD, T.R., BASKIN, D.S., EISENBERG, H.M., FLAMM, E., LEO-SUMMERS, L., MAROON, J., MARSHALL, L.F., PEROT, PL., PIEPMEIER, J., SONNTAG, V.K.H., WAGNER, F.C., WILBERGER, J.E., WINN, H.R., and YOUNG, W. (1992). Methylprednisolone naloxone in the treatment of acute spinal cord injury: One year follow-up data. Results of the second National Acute Spinal Cord Injury Study. J. Neurosurg. 76, 23-31. GEISLER, F.H., DORSEY, F.C., and COLEMAN, W.P. ( 1991 ). Recovery of motor function after spinal cord injury: A randomized, placebo-controlled trial with GM-1 ganglioside. N. Engl. J. Med. 324, 1829-1838. or
Address reprint requests to: Wise Young, M.D., Ph.D.
Department of Neurosurgery University Medical School
New York
550 First Avenue New York, NY 10016
DISCUSSION Dr. Wrathall: What is the influence of the red blood cells that are trapped? They replace neural tissue, and I presume are counted as "intact tissue." Dr. Young: Infiltration of RBCs would contribute to the sodium minus potassium difference so that there would appear to be intact tissue. This measurement allows you to distinguish only the cellular volume fraction. This is very straightforward. The RBCs also will never reach beyond the hematocrit of blood, which in a rat is typically about 35%. I have never seen the cell volume fraction at the impact site go below 30%. Maybe it is because of blood entering the tissue. Dr. Wrathall: I must admit that we have been disappointed in the correlation between behavioral or electrophysiologic measurements and tissue damage, and we have looked at many spinal cords. But I do not trust our histologie measurement of the tissue sufficiently so that I can show a graph relating sodium minus potassium and histology. This is a destructive technique, and once we kill the animal for this measurement, we 153
YOUNG cannot then look at the
same
spinal cord 3
weeks later for
histologie changes.
It is my
general impression,
however, that it predicts the degree of histologically apparent damage at 2 or 3 weeks after injury and certainly
predicts our 3 or 4 week injury severity levels.
Dr. Tator: The increase in total sodium at the injury site is somehow balanced by a major increase in calcium intracellularly. If one were to examine only sodium minus potassium as the index of injury and treatment, are we going to miss something? Can we see effects for drugs that act primarily at the level of calcium? Dr. Young: Of course. If a drug saves cells, it would show up in the sodium minus potassium difference. Na minus K directly reflects the volume of cells that can maintain their sodium or potassium gradients across a membrane. By the way, the amount of calcium is very small compared to sodium and potassium, about 1 or 2mM. Dr. Tator: That's why I am concerned about looking at only sodium and potassium. If the drug acts on calcium and you don't need huge amounts of calcium inside vs outside to kill cells, then what? Dr. Young: What we are essentially measuring is the hematocrit of the tissue. We are using sodium minus potassium as a measurement of the cell volume fraction, and we have derived this equation from one simple assumption. As long as we assume isotonicity (Na K) is proportional to Vi/Vt and Vi/Vt is the cell volume fraction of the tissue, we are saying that if you hit the spinal cord, there will be loss of cells. The cells will join the extracellular space, and if you have any treatment that saves those cells, you have less change in (Na K). It does not address mechanisms. It is a simple measure of cells that are intact in the tissue. Dr. Stokes: We are dealing with an extremely heterogeneous tissue, and the cells we are interested in may not be the ones that are showing these dramatic shifts. Intuitively, getting an index of damage is important, but I think it must be closely tied to relative histopathologic index in specific cell types. Dr. Young: No. You certainly do not do that in your laboratory when you look at necrosis. I am telling you that if you are looking for necrosis in the spinal cord and it takes you 6 weeks to do 10 animals to map out the necrotic area, you can do that in 1 day using the sodium minus potassium method. Dr. Stokes: You don't agree that this index would have to be correlated with histopathology, axonal types, or neurons? Dr. Young: That's why we do electrophysiology. This method is a simple, straightforward, robust measurement of cell volume fraction. It is equivalent to doing serial sections of the tissue, mapping out the necrotic area straight through, and then summing it up. I presented a slide showing a direct linear correlation between the necrotic area in the tissue and the sodium minus potassium changes. In fact, it predicts precisely both the extracellular ionic differences and the expected gradient across the membranes. Dr. Hsu: We have been looking into vasogenic edema in the brain and spinal cord. Evolution of edema in the spinal cord after trauma is similar to your sodium minus potassium results. Maybe you have developed another way to determine vasogenic edema. Dr. Young: We have measured albumin entry into the tissue and correlated with sodium minus potassium, and indeed there is a nice correlation. I think that necrosis clearly correlates with the degree of endothelial breakdown in the tissue. But the nice thing about sodium minus potassium is that this is a relationship that predicts precisely the volume of intact cells. Dr. Faden: The highest predictor in terms of correlation coefficient (around 0.96) with regard to outcome and injury severity is tissue magnesium. Tissue magnesium also correlates highly to lipid hydrolysis, so there is both a mechanism and a predictor that seem to be very sensitive to drug treatment effects and to degree of injury. I wonder if you could comment on the magnesium component and whether you think there might be an avenue for using both the sodium minus potassium and magnesium together as a combined predictor. Dr. Young: We measure magnesium in all our tissue specimens. It does correlate extremely well with the sodium minus potassium. I believe magnesium reflects the amount of lipid loss in the tissue, and lipid loss is one of the primary markers of necrosis. By using extracellular ion selective microelectrodes in the rat spinal cord injury model, we have found that the rat clears out the sodium and potassium very rapidly from extracellular space. Extracellular potassium is normalized by 20 min after an impact injury, even in severe injuries. I believe that you can assume that sodium minus potassium in the extracellular space is equal to plasma levels by 30 min after injury, and from that point, you can simply analyze the data. I must put in a caveat, and that is we are making the cellular volume fraction calculations based on the value of the ion —
—
154
END POINT MEASURES: BIOCHEMICAL
gradient across membranes (G), and there is no easy way of determining this. As it is turning out, the ionic gradient probably is low for as long as 6 h after injury. We always take a segment of cord that's far away from the injury site to estimate a G value, and we also take specimens from other parts of the body to be sure that we do not have an animal with a G value that is remarkably different. We are still working on finding a method where we can use the G value from a normal piece of tissue to normalize our data at the injury site so we can apply earlier analyses.
155
