Incrementally Applied Multifaceted Therapeutic Bundles in Neuroprotection Clinical Trials…Time for Change
- First Online:
- Cite this article as:
- Kofke, W.A. Neurocrit Care (2010) 12: 438. doi:10.1007/s12028-010-9332-7
- 189 Views
We have reached a stage at which research in this area should stop altogether or radical new approaches adopted.
Targets of acute stroke therapy trials 
Drugs to improve blood flow
Drugs to protect brain tissue
Calcium channel blockers
Free radical scavengers—antioxidants
Improve capillary flow
Sodium channel blockers
Potassium channel openers
Mechanism unknown or uncertain
Leukocyte adhesion inhibitors
Nitric oxide inhibitors
Suggest an explanatory complex biological system model for such reproducible futility thus providing a rationale to suggest that such negative results, in retrospect, are an expected outcome,
Provide support for the notion that only a multimodal approach can reasonably be expected to lead to effective neuroprotective therapies, and
Propose a practical research paradigm to develop a multimodal approach which, taken altogether is more likely to produce breakthrough level results that will then be suitable for subsequent multi-institutional evaluation and generalizable implementation.
Moreover, the concepts presented here in the context of neuroprotection, are probably generalizable and applicable to many other complex clinical problems. Certainly, the efficacious use of combination therapy already employed in management of cancer  and AIDS  provides examples of this approach that might be adapted to neuroprotection.
Impact of Perturbations in Complex Systems
Single Facet Futility—Why Negative Results, in Retrospect, Are an Expected Outcome
Further adding to the complexity of this pathobiological process, the weighting variable for each factor changes with time after the initial insult, because the importance of various secondary processes wax and wane. For example, the weighting factor for cerebral blood flow impact on outcome severity can be very high early on but becomes less important with time as the weighting factor for reperfusion pathophysiology, initially low, increases with the time after ischemic injury.
Much of the knowledge about pathophysiology and therapy after brain injury is derived from animal models…designed to reduce biological variation, and without complicating concomitant illnesses and pharmacotherapy. Humans, however, present with myriad additional confounding biological and pharmacological factors not included in animal models. In addition, in the clinical environment there are associated system factors, Hi, such as nursing ratio and experience, drug and technology availability, day of the week and time of day, rapid response team availability and quality, varied standard clinical practices , and so on, which also can be important to the eventual injury outcome and likely also lack linearity  and with time-dependent variation . Reports of therapeutic success with single institution studies followed by failure in multi-institutional studies, e.g., hypothermia used in traumatic brain injury, is an example of the important impact of system factors and their variation in care [5, 8].
Since there are multiple highly variable and interacting linear and non-linear biological and health system factors that enter into a given outcome, it should come as no surprise that clinical studies designed to ameliorate only one of many complex factors tend to show no or limited therapeutic effect, especially when multi-institutional in design (increasing variation in H). The exception is when there is truly a breakthrough phenomenon (large W, e.g., early thrombolysis in ischemic stroke [9, 10]) or the therapy exerts an intrinsic multifaceted effect (e.g., hypothermia after cardiac arrest [11, 12]).
Pathological perturbations of homeostatic mechanisms are not easily corrected by single facet corrective measures in the context of biological complexity.
This leads to the concept that there are serious limitations in the generally accepted use and interpretation of prospective randomized monotherapy studies in the treatment of complex diseases like stroke and TBI such that the medical community has become intellectually paralyzed by traditional scientific notions of knowledge creation. Furthermore, the current widely accepted methods of advancing clinical knowledge for complex problems through single-hypothesis-oriented clinical research may be an unnecessary waste of public resources that has produced innovation paralysis on the part of institutions, third party payers, clinicians, pharma , and investigators . An alternate research method that is based on a multifactorial approach to introduce therapeutic bundles which, in aggregate, can be subjected to testing therefore is needed.
Rogalewski et al.  and Hemmen and Lyden  recently reviewed and endorsed the concept of multifactorial therapy in neuroprotection. Moreover, many authors have demonstrated the feasibility of this approach in creating dramatic levels of neuroprotection in preclinical studies [16–19] and, undoubtedly the multifaceted mechanisms attributed to hypothermia  account for its successful translation in cardiac arrest [11, 12] (notably, based on single center studies). None of these authors, however, suggest a rational means of how to introduce the multimodal approach other than trying everything at once…perhaps another prescription for research failure.
Multimodal therapy is effective in preclinical neuroprotection research and in other diseases. A rational method is needed to develop multimodal therapeutic bundles in neuroprotection.
PDSA (Plan-Do-Study-Act) Cycles—From QI to Generalizability
The so-called PDSA method advocated by Berwick  for local quality improvement may provide a technique to develop an incrementally implemented multifaceted approach to serially test and add single clinically unproven but safe, pathophysiologically sensible, and otherwise scientifically supported facets in the therapy of a complex disease. Such an approach should produce a multi faceted therapeutic bundle that then may be amenable to more widespread testing and/or application.
The PDSA method usually entails application of a nonrandomized process, using institutional and individual (paired or N of 1) historical control data, to introduce incremental improvements in processes of care. The first step, planning, entails identification of a process to improve with a plan for implementation. The do phase entails the actual systematic implementation of a new process of care. Studying then is the procedure for collecting and analyzing the results of the new intervention. The act phase entails coming to a conclusion that the results have merit or not and then to produce a permanent yet incremental change in a system’s health care protocols.
The PDSA technique is designed for processes of therapeutic QI that are meant to improve process-of-care outcomes locally and not necessarily produce generalizable knowledge. Thus, by the federal definition of the United States such processes technically are not research. Nonetheless, the PDSA process is a widely accepted type of institutional process-of-care experimentation which very well could be used as a method to produce or contribute to the production of new generalizable knowledge for complex biological diseases.
Using PDSA Cycles to Develop a Multifaceted Program to Arrest Secondary Brain Damage
Impact on the target pathophysiologic pathway.
Surrogate neurologic outcome measures.
Functional neurologic measures.
- Impaired ATP production during resuscitation and after return of spontaneous circulation
- Near infra-red transcranial laser light therapy 
Measure ATP and phosphocreatine with magnetic resonance spectroscopy
- Reperfusion related free radical injury
- NXY-059 
Measure plasma and urine isoprostanes
- Increased intracellular calcium
- Dantrolene 
Measure plasma dantrolene concentration
- Mitochondrial permeability transition pore opening
- Cyclosporine 
Measure plasma or CSF cyclosporine concentration
- Glutamate associated injury
- Ketamine and barbiturate infusion 
Measure plasma ketamine and barbiturate concentrations
EEG to evaluate brain effect
Evidence supporting the utility of this approach may be derived from observations that implementation of the “therapeutic bundle” advocated by the Brain Trauma Foundation is associated with a dramatic improvement in outcome after traumatic brain injury . Notably, the individual facets neither have simultaneously developed nor have they been devised in a coordinated manner. This is said to be evidence based; however, only one of the 14 recommendations is based on level one evidence and five are based on level three or less evidence. This contrasts with the relatively unchanging morbidity and mortality over recent decades [36, 37] after intracerebral hemorrhage, a disease which has not been subjected to the same multifaceted therapy as has occurred with traumatic brain injury.
The PDSA method works well to serially and incrementally improve local processes of care and should be adaptable as a method to develop generalizable multimodal therapeutic bundles for neuroprotection.
It may be impossible to obtain timely information regarding the pathophysiologic efficacy of a given therapy. For example, a drug directed at altering a brain neurotransmitter would require invasive brain microdialysis data to really show the effect. This may not be ethically justifiable.
It is unclear the best way, statistically to show pathophysiologic and surrogate outcome efficacy. One issue, as presented, is the use of internal historical controls. Such an approach introduces the well-known possibility of bias due to unmeasured unknown processes that may systematically change over time within an institution. An alternate design would call for an untreated control group which would undergo the same surrogate and functional outcome measures. Given that the intent is simply to efficiently create a therapeutic bundle for later comparison with an untreated control group, this approach may not be necessary, but nonetheless certainly a reasonable alternative.
This model, as currently constructed, provides for no appropriate dose-finding methodology. This is a relatively new and developing area in design of clinical research studies and, conceptually, such techniques might be included in the PDSA approach, but using pathophysiologic endpoints, where available, as the goal in dose finding.
Sample size calculation may be problematic. Pathophysiologic endpoints probably could be used with serial assessment during the study for the likelihood and minimization of type I and II errors. By the end of therapeutic bundle development there should be enough preliminary data available on clinically meaningful outcomes to then be able to estimate sample sizes for a subsequent properly conducted prospective randomized trial of the therapeutic bundle versus an untreated control group.
Interactions between therapies may arise. This may be unavoidable and hopefully the interactions will both be in a positive direction, although this may not necessarily arise. It seems the best approach would be to monitor for interactions by continually measuring pathophysiologic endpoints for previously accepted therapies. For example, inclusion of an antioxidant in the accepted paradigm should then mandate continued monitoring for isoprostanes as new therapies are added.
It is unclear how many modes to test within a given therapeutic bundle nor the order in which to serially evaluate them. Ideally modes are serially added until a statistically significant clinically meaningful effect is observed. Moreover, given a desire for efficient use of resources, the planners should seek to test and implement modes earlier that are expected to have higher impact. However, translation to testing in a broader multi-institutional context will increase heterogeneity such that a higher level of statistical significance may be first sought before going to a prospective randomized multi-institutional evaluation of the therapeutic bundle.
Finally, is it really necessary to introduce each facet singly? Perhaps bundles that are directed to a single goal could be added simultaneously in a goal directed manner .
By the end of the development of a given overall therapeutic bundle, multiple new neuroprotective therapies (versus secondary injury) will have been added to the neuroprotective cocktail as part of an IRB overseen PDSA process. A robust difference in surrogate and functional outcomes should be apparent compared with historical controls. The results can then be offered in aggregate as a multifaceted therapeutic bundle ready for multi-institutional trial or…for a true breakthrough effect…possibly for immediate generalized implementation.
Although not satisfying in a purely scientific sense, this approach may nonetheless offer the potential to introduce effective breakthrough therapy for acute neurologic syndromes that heretofore have proven recalcitrant to solutions using standard scientific methods. As such, the potential benefits seem to outweigh possible risks.
This was presented at the 2008 meeting of the Society of Neurologic Anesthesia and Critical Care and the 2009 meeting of the Neurocritical Care Society. The author gratefully acknowledges the helpful comments of Peter LeRoux and Sherman Stein in the development of the manuscript. The author has no conflicts of interest.