Two major studies in the field of therapeutic hypothermia—the National Acute Brain Injury Study: Hypothermia I and II—have shown the high safety levels of body cooling administered very soon after a severe brain injury. However, there are still questions on how best to forge ahead. How should we be combating the major complications of low blood pressure and rebound intracranial pressure during rewarming? Dr. Guy Clifton, the principal investigator in the NABISH studies, leads an expert panel discussion on addressing these issues, as well as what the future holds for rewarming methods.
Dr. Clifton: In the second hypothermia study that this group performed—the National Acute Brain Injury Study: Hypothermia II (NABISH II)—we found a rebound increase of elevated intracranial pressure (ICP) during rewarming that had a detrimental effect on outcome and mortality rate in patients with diffuse brain injury. We did not see such an effect in the first study. So, the first question would be, why the difference?
Dr. Okonkwo: We took all forms of severe traumatic brain injury (TBI) and we applied a common paradigm of hypothermia to all of those groups. But as we get more mature in our approach to TBI, we understand that there is a tremendous amount of heterogeneity within the single concept of “severe TBI,” and therefore, if you have patients who have diffuse cerebral edema or diffuse brain injury and you compare that against a separate cohort of patients that has intracranial hematomas, well, it is very common for patients with intracranial hematomas to have early raised intracranial pressure. In fact, that is why many patients in that particular cohort require early craniotomies, either immediately after the resuscitation phase or shortly after their admission to the hospital. If you contrast that to patients that have diffuse brain injuries, their peak period of raised intracranial pressure tends to be between 48 and 72 hours later.
Thus, in the NABISH study, we were in fact rewarming patients with diffuse brain injuries coincident to their peak phase of intracranial hypertension. So it is possible that we were simply either observing or potentially exacerbating the natural history of patients with diffuse brain injury by rewarming them at 48 hours: observing the natural history because that is the start of the peak phase of raised intracranial pressure, or exacerbating the natural history because we know that rewarming carries with it a risk of rebound ICP. So you may take patients who were on the edge of intracranial hypertension and push them over the edge by rewarming them right at that critical moment.
Dr. Clifton: Good comment. Alex?
Dr. Valadka: The short answer is, I do not know, and I cannot explain. David may be right in his previous comment. If you look at some of the most important differences in the way the two NABISH studies were carried out, in the second study the patients were cooled quicker—over around 4 hours as opposed to 8 hours—but I do not think that can explain the higher rate of rebound hypertension.
Another difference was the fluid balance. More fluid was given in the second study, which was a good thing because it prevented the hypotension that had been seen in the first study. But again, whether you could somehow use that to invoke an explanation for the higher ICPs, I do not know, especially if the serum sodium osmolality did not seem to be affected by the extra fluid.
Dr. Clifton: The two differences in the study that were most clear were fluid balance and the rate of hypotension during the period of hypothermia. First, I'll address the mean fluid balance in the hypothermia group. In the first study, at 96 hours, patients were 3 liters positive; in the second study, they were 5 liters positive. In the normothermia group, they were 2 liters positive at 96 hours; and in the normothermia group in the second study, they were 3 liters positive. Now, the other difference is that the hourly dose of morphine in the second study was close to nearly half of what it was in the first study. We used far less analgesia.
The rate of hypotension during the period of hypothermia, 96 hours, in the second study was much less than it was in the first study. That is, mean arterial pressure was less than 70 millimeters of mercury (mmHg). So my interpretation of it has been that the measures that we use to abort hypothermia-induced hypotension produced rebound increased ICP, at least in patients with diffuse brain injury.
So, given that you want to avoid hypotension, we appear to have successfully done that by attention to use of vasopressors, which was equivalent in the two studies, but with more fluid expansion and less morphine in the second study. Given that this combination reduced hypotension, how might we alter the rewarming protocol to both avoid hypotension and avoid rebound increased ICP? How might we change things?
Dr. Valadka: Guy, I am wondering if this is echoing some of what our colleagues in Lund, Sweden, have been saying. Is it possible that all the emphasis on maintaining a higher blood pressure in the second study somehow did contribute to a higher ICP? The flip side is that a lower blood pressure in the first study may have decreased the pressure inside the arteries in the brain and possibly contributed to a lower ICP.
Dr. Clifton: No, but (there is) the distinct possibility. Actually, cerebral perfusion pressures (CPPs) were not driven to very high levels. What we did not see is a lot of patients with CPPs of 80 and higher. We just found fewer patients with CPPs below 60, and we do not have a lot of evidence that we drive CPPs very high. So it could still occur in the absence of that measure, I think.
Dr. Aisuku: Was there, in the second study, a greater use of vasopressors?
Dr. Clifton: The percentage of patients that had vasopressors and had two vasopressors was almost identical in the two studies.
Dr. Aisuku: Okay, so less analgesia, more fluids, and about the same amount of vasopressors—
Dr. Clifton: Yes.
Dr. Aisuku: —led us to a higher percentage of patients with increased rebound ICP?
Dr. Clifton: Yes. I would say, too, Imo, this is a hard thing to quantify, but there was a lot of discussion and far more emphasis on maintaining blood pressure during rewarming in the second study than in the first study. There were far more centers in the first study and, therefore, adherence to the protocol was much better in the second study than the first.
You cannot measure the dose of vasopressors. The percentage of patients that had them was the same, but we cannot speak to how aggressively they were administered.
Dr. Aisuku: I see.
Dr. Okonkwo: I think it is important to recall that the intent of the hypothermia treatment—in both NABISH trials—was targeting the underlying pathophysiologic processes ongoing in the injured brain and injured neurons, as opposed to a treatment strategy that was meant specifically to address intracranial hypertension.
Dr. Clifton: Correct.
Dr. Okonkwo: We know from many studies over the last 20-plus years that if you look at the preponderance of evidence—there have been some studies that have contradicted this—it supports hypothermia as a very effective treatment for intracranial hypertension, as opposed to the concept of a primary treatment of the underlying pathophysiology. So perhaps in the NABISH trial, what we needed was to deepen hypothermia to 33 degrees for that critical time; early induction of hypothermia, getting them down to 33 degrees quickly, and then maintaining that hypothermia of 33 degrees for 48 hours to target the primary pathophysiology. But rather than rewarming them at the 48-hour mark, there was very likely a cohort of patients that would have benefited from either maintenance of hypothermia at 33 degrees or only partial rewarming to 35 degrees until the risk phase of intracranial hypertension receded.
Dr. Aisuku: Agreed.
Dr. Clifton: I think in retrospect, I agree with that statement. Alex, what is your thought on this?
Dr. Valadka: I agree. As we were saying earlier, the analogy is very applicable to other ways we treat high ICP. So if you are going to give mannitol or hypertonic saline or drain cerebrospinal fluid (CSF) or use a barbiturate-induced coma, you do not just, say, only give it for 48 hours or a certain amount of time. You administer those treatments for as long as the patient needs it.
Dr. Clifton: So I think the summary of our collective judgment is not that we necessarily change the approach to fluids and analgesia if we want to avoid hypotension, but that we individualize the rewarming to the response of the patient and the patient's ICP?
Dr. Aisuku: Correct, because I think, in response to your second question, I still think normovolemia is the optimal goal, and avoiding hypotension is an optimal goal as well, whether it is some combination of fluids and pressors, etc. So the presumption is, if you feel they are hypovolemic, you should resuscitate with fluids until they are normovolemic, and then employ some degree of pressors so you do not go overboard.
During rewarming, your focus needs to shift as you de-escalate some of these measures. If you continue hypothermia, I would agree that watching the ICPs over those 48 hours should guide the rewarming rate—and I would honestly extend it out to 7 days if ICPs remained elevated.
Dr. Valadka: I agree with what Imo said, and I would like to follow up on some of Guy's experience from when he was first putting these protocols together that concerns profound and prolonged hypothermia for things like infection and coagulopathy. As Guy said earlier in this conversation, he needs to individualize the treatment. So even though you may prefer to keep a patient cool for several days, if they are developing a worsening pneumonia or other infection, or if the coagulation profile is going up or platelets are going down, that is when you have to make individual judgment to rewarm the patient sooner to prevent systemic complications.
Dr. Clifton: You know, another thing we never knew was the depth of hypothermia. You could surmise that if you did not take a patient to as low a temperature that you would not have as much rebound when you rewarmed. The reason for 33 degrees is it was the lowest safe level, and, in the laboratory, there were a couple of experiments with a graded response. That is, you had better neurologic outcomes in experimental animals at 33 degrees than at 35 degrees. But who knows what that translates to in terms of patients?
Another aspect of individualizing management would be to rewarm a couple of degrees to 35 and extend the amount of time at 35 degrees where you would have fewer complications. So depending upon the patient's physiologic state and so forth, you might be able to compromise with a lower temperature.
Dr. Okonkwo: Guy, you have some data that suggested one of the keys to rewarming was the time to 35 degrees?
Dr. Clifton: Right.
Dr. Okonkwo: That, in a way, it was more important to get to 35 degrees very quickly?
Dr. Clifton: That is correct. For hematoma patients, that was the number. It was overwhelmingly 35 degrees and the time to 35. Now, for diffuse brain injury, there was really no effect—early or late cooling in diffuse brain injury in either study that was detectable at all.
Dr. Okonkwo: That probably goes back to the concept that the maximum efficacy of hypothermia in the diffuse brain injury patients may in fact be between 48 and 96 hours, when those patients are at highest risk for their underlying pathophysiologic processes reaching their worst.
Dr. Clifton: If I had to guess, it would be that the application of hypothermia in patients with diffuse brain injury is a very good means of ICP control. What we saw when we looked at the hematoma data was if you got to 35 degrees within 1.5 hours of surgery start time in both studies, there was a big effect.
Dr. Okonkwo: Excuse me, just to clarify my question: you mentioned that for patients with diffuse injury, hypothermia can control ICP, but I think one of the questions that led to the study was whether it can actually be a neuroprotectant. So do you think that in the patients with hematoma, the hypothermia may have a more neuroprotectant effect as opposed to simply treating the secondary problem of high ICP?
Dr. Clifton: My hypothesis is that we used hypothermia as a neuroprotectant in both studies. However, in patients with diffuse brain injury in the second study, the measures we used to abort hypotension came at the expense of rebound ICP. Because of the timing of the rewarming, as David Okonkwo has pointed out, during the peak period of ICP they were actually worse. Japanese investigators have done large studies of patients with diffuse brain injury with normal ICPs treated with hypothermia that showed no (neuroprotectant) effect at all and by careful investigators—nothing. So I doubt that hypothermia has a neuroprotective effect in patients with diffuse brain injury.
But back to patients with hematomas, which is a really different group of patients, what we saw in both studies was that if we hit 35 degrees within 1.5 hours of the start of surgery—which would have been before reperfusion, before the hematoma was removed, and before the brain reperfused—there was a big effect on the outcome, with relatively little rebound ICP. There was a little (rebound), but it was muted, and certainly nothing like we saw in diffuse brain injury. So it was a completely different pattern.
My interpretation of that finding was that the reason rewarming was less of an issue there is due to the protection of the brain during reperfusion, which I believe is done in cardiac arrest and neonatal hypoxia. Ischemia with reperfusion does not really occur in diffuse brain injury. So I believe it is completely different pathophysiologies, and probably calls for completely different protocols for both cooling and rewarming.
Dr. Okonkwo: In essence, the patient who has a diffuse brain injury and the patient who has an intracranial hematoma requiring surgery are not suffering from the same disease, even though we label each of those patients as a “severe TBI patient.”
Dr. Clifton: I think if we have demonstrated nothing else, we have demonstrated that the treatments need to be different.
Dr. Aisuku: So that would beg the question, then, for the hematoma patient, is it adequate to take them to 35 degrees and maintain this temperature? And then, for the diffuse injury, although we all agree that they have different pathophysiology, we just do not know if treating ICP for a greater period of time would have any beneficial effect.
Dr. Okonkwo: I would be very curious to look back at the primary data from the Chinese studies, where the hypothermia paradigm was to induce hypothermia and maintain hypothermia until the ICP is normalized. So you wonder if the reason why the Chinese studies demonstrated a treatment effect was because they had more hematoma patients than diffuse brain injury patients in those two studies.
Dr. Clifton: I know one of the lead investigators—his name is Jiang, in Shanghai—and he is very careful. I absolutely believe his data. What he did is to cool after about 8 hours, and sometimes later. They were in no hurry to cool. But, as David has pointed out, in this study, investigators maintained hypothermia until ICP elevations diminished. They found that hypothermia-treated patients had much-improved outcomes. I do not know the exact distribution, but from the timing, these patients with evacuated hematomas would have been cooled well after hematoma removal. I believe that three-quarters or more of the patients were diffuse brain injuries. The findings in that study are consistent with the hypothesis that the treatment effect was from reduction of elevated intracranial pressure.
Imo, to respond to your question, we do not know if 35 degrees is adequate or 33 degrees in patients with hematomas. Lab data are so strong, and cardiac arrest data are actually quite strong for 33 degrees. So I am going to assume that if the treatment effect we found in patients with evacuated hematomas is the same mechanism as in cardiac arrest, patients are better off at 33 degrees. But I would certainly tailor the rewarming either way to the patient's ICP. To me, that seems like a sensible thing to do. I wish we would have done it, but from the first study there was no indication that rewarming at 48 hours was a problem.
Dr. Aisuku: That would make sense. I think it is hard to do a dose–response type of evaluation in this and go back and (say which temperature) 33, 34, or 35 or whatever the case may be. From complications that we were concerned about regarding prolonged hypothermia, anecdotally there is a difference at 33 than at 35 from pneumonia and other complications. So if you had to maintain them and could maintain them at 35, I think you have a much better chance of not having to actively worry or manage the other primary concerns.
Dr. Clifton: I agree. With hematoma patients, cool them to 33 as fast as you can—understanding that you have to get to 35 before the hematoma comes out to get an effect—maintain 33 for a day, possibly two, and then 35 thereafter, depending on the response of the patient, would be, to me, a pretty good formula. It would be interesting to think about how you write a protocol in which you deliberately individualize management and sell that to a study section. It is doable, but I am thinking of the flow diagram.
Dr. Aisuku: Very good question.
Dr. Clifton: Of course, that is a political question, not a clinical question.
Dr. Aisuku: Of course, our conversation is occurring after we had a very interesting case, in which we were employing hypothermia for ICP management, and as part of the protocol, they were also put in a pentobarbital coma, and the patient's potassium dropped to 1.5 during the cooling process, and then swung to significant hypokalemia during the process, which sparked a lot of debate. Part of the debate was how much of that was driven by the pentobarb. Mind you, I will add a caveat; they were given double the recommended dose of pentobarb, which is another story. But how much of that was pentobarb-driven and how much of that was hypothermia-driven? I think just as part of a general discussion, the electrolyte components of it, other than this one experience, I think in general the hypokalemia that I have seen with them is relatively easily managed. Below 3.5, I treat very mildly and mark it for when they rewarm, just being aggressive in how frequently I monitor the potassium, knowing that their total body potassium is not necessarily effectively changed.
Dr. Clifton: Would you like to know where the very strong statement we made in the protocols not to administer either magnesium or potassium within 8 hours before rewarming came from?
Dr. Aisuku: I would be very interested.
Dr. Clifton: It was not from a systematic study. We knew from all of our work that potassium went up during rewarming; there was no question, as intracellular stores are mobilized and pushed into the extracellular compartment. So I cautioned everyone about this. In the first study, one day on a Sunday, I got a call from a center that had not enrolled many patients, and the patient in question had a potassium rate of around 2, clearly hypokalemic.
They were ready to rewarm, and the attending physician was insistent upon giving 40 or 60 milliequivalents of potassium before the rewarming. I said, “Do not give any at all. You do not want to do this.” He did, and the patient coded with a potassium of something around 7. I did not have to have that conversation twice with anyone.
Dr. Aisuku: I think some of my staff have learned that lesson the unfortunate way.
Dr. Valadka: But Imo, to clarify, so that patient you are talking about was given potassium?
Dr. Aisuku: Yes. So here was the challenge. I think they were given potassium at the 1.5 level, which I do agree with. But they continued to give potassium despite that, and then began actively rewarming while they were replacing the potassium.
Dr. Clifton: The problem with it is everybody is different, and you just do not have any way of knowing how much potassium these patients are going to mobilize. It is probably dependent upon muscle mass, things that are difficult to quantify if that is a factor. It is certainly dependent upon rewarming rate. In all of the studies, there is no evidence that there is a higher rate of ventricular arrhythmias or any cardiac arrhythmia, atrial or ventricular, in hypothermia-treated patients than others. They actually do seem to tolerate hypokalemia during hypothermia pretty well.
Dr. Valadka: You mentioned there was some concern about the pentobarb causing hyperkalemia. That is very, very rare.
Dr. Aisuku: I completely agree. When the case came across my desk to review, I think part of it was the group who got involved in the care of the patient had not done a lot of hypothermia management in the past, and so did not have a clear way of understanding how low the potassium could potentially go, and was looking for other potential causes, because they did not believe that hypothermia alone could drive the potassium that low.
Dr. Clifton: Oh, it can.
Dr. Aisuku: Exactly. So that was where that end of the discussion came about, and in my conversation with them, yes, it certainly could, and they just needed to be careful regarding that, which is a completely separate issue from the pentobarb. But it was an interesting discussion, so I included it as part of what had gone on from my end.
Dr. Clifton: So I think our conclusions are that the rate of rewarming and the duration of hypothermia, and perhaps the level of hypothermia during rewarming, after cooling, should be dependent upon the clinical condition of the patient, particularly ICP, and systemic complications that might be exacerbated by hypothermia.
Dr. Valadka: I might even go a little bit further and say that in the experience of those of us on this call, the adverse effects of hypothermia are reasonable when hypothermia is done in a good ICU. Even though in the past we have limited the duration of hypothermia out of concern for the side effects, I think it is very reasonable to say that the side effects are manageable. On the other hand, if we rewarm too early, the ICP can get out of control. So I think there needs to be a reassessment of the risk–benefit ratio. I think the risks of a longer period of hypothermia are not as great as had been feared several years ago, but the benefits may be substantial.
Dr. Clifton: I think we have learned that the systemic complications of hypothermia are far less than we feared. There is just no evidence at all of a significant increase in delayed hemorrhage in patients with intracranial hemorrhages treated with hypothermia. Within the parameters we have used, the electrolyte abnormalities are predictable—the result of hypothermia. And there is really not, for a 48-hour duration at 33 degrees, much evidence of increased rates of infection. So I agree.
Dr. Okonkwo: I think that the safety of hypothermia is the strongest lesson learned from both NABISH I and NABISH II. The disappointing part of things is that the concept of hypothermia and the utility of hypothermia may get lost in the disappointing results of the trials. While those two studies clearly demonstrated the safety of hypothermia as a concept, we have not been able to translate the fact that hypothermia has protected every rodent ever exposed to it, but we have not been able to translate that concept into human TBI. The reality is that hypothermia for the treatment of human TBI works for certain patients, and we have just not been able to demonstrate in a robust and statistically significant fashion that hypothermia should be used under “these” circumstances to “this” degree for “this” duration in “this” group of patients, and to a different degree for a different duration in other groups of TBI patients.
Dr. Valadka: But while we have not done the Phase III study that describes the protocol I am going to mention, I think that in aggregate we can conclude that the real question is in terms of management. We know that the major complications of hypothermia and brain injury patients are hypotension and rebound ICP, depending upon the timing and rate of rewarming. We know that hypothermia is safe if you control hypotension, meaning vasopressors, fluid expansion, and low doses of analgesics. We have concluded that you want to individualize; we believe collectively that you should individualize the rewarming rate in relation to the ICP of the patient.
And we surmise from the data of a Phase III study—that looked at patients with hematomas—the way to induce hypothermia is to reach 35 degrees within 1.5 hours of surgery as soon as possible but before removal of the hematoma. In diffuse brain injury, if it is applicable, (hypothermia) appears to be a means of ICP control, with some question of whether it could have a neuroprotectant effect if rendered early.
So there are significant questions. But we do know you have to control hypotension. We know how to do it, and we know it means you have to individualize rewarming.