Tuesday, February 14, 2017

Microelectrode Recordings: How Much is Enough

GNC Professional Newsletter
Volume 3, No. 2
February 14, 2017
Erwin B. Montgomery Jr. MD

Microelectrode Recordings (MERs)
It would appear obvious that there is a great diversity in surgical approaches to Deep Brain Stimulation (DBS) lead implantation.  In the near mathematical impossibility of randomized control trials that pits each different approach against the  others, how is one to decide which approach or method is best?  These newsletters have argued that even in the absence of Evidence Based Medicine (EBM), the surgeon still has the responsibility of doing what is reasonable.  Surely, few would argue — until an attempt is made to define what is reasonable.
Past newsletters have framed the issue as being between MERs and  image-based targeting.  Further, the MER issue was couched in terms of single or multiple microelectrodes.  However, recent experience suggests that the diversity is far greater than the previous framing would suggest.  As alternative methods are encountered, these newsletters will attempt to offer a critical analysis.
Some neurosurgeons and intraoperative neurophysiologists move the microelectrode to the image-based target, for example, the ventral extent of the ventral intermediate nucleus of the thalamus (Vim), subthalamic nucleus (STN) and the globus pallidus interna (GPi).  MERs are conducted over few mm’s.  If the neuronal recordings are consistent with expectation, the DBS lead is implanted.  Assuming that neurons are sampled over a couple of mm’s, then this may not be the most reasonable approach. 
In the case of the STN, a consensus criterion for optimal DBS location is a trajectory that demonstrates at least 5 mm of sensorimotor-related neurons.  This consensus was based on anatomical considerations that the widest extent of the STN in a typical approach is 5 to 7 mm.  Thus, any movement of the trajectory from a prior trajectory that contains at least 5 mm of STN is not likely to pass through most of the sensorimotor STN.  This means that at the minimum, MERs should commence at 5 mm above the image-based target.  Past studies have demonstrated that the variability of the physiologically defined optimal location in the axis of the trajectory is very large, on the order of 7 mm (Montgomery EB Jr., Microelectrode targeting of the subthalamic nucleus for deep brain stimulation surgery. Mov Disord. 2012 Sep 15;27(11):1387-91. doi: 10.1002/mds.25000. PubMed PMID: 22508394).  This means that starting only 5 mm or less above the image-based target has a high probability of not meeting the criterion. 
An important corollary is that the criterion for STN can establish whether or not the trajectory is optimal for DBS lead implantation.  The greater challenge is determining the direction and extent of the movement to the subsequent MER or DBS lead trajectory.  Important information is obtained from the entire trajectory over 15 to 20 mm.  For example, the depth at the bottom of the thalamus and the depth at which the STN is entered and exited provide important clues as to where the more optimal trajectory might be.

For Vim, the criterion is to find the proper homuncular representation in Vim, which most often is the distal upper extremity.  Further, the head region should be avoided to prevent increased dysphagia, dysarthria and aphasia with DBS.  However, the homuncular representation in Vim is complex.  The representation has a layered structure, much like an onion.  Thus, in a trajectory that is too medial one still may encounter neurons related to sensorimotor testing of the upper extremity.  If only a small segment of the trajectory is studied, then there is a chance that the head representation would not be encountered, increasing the risk of implanting the DBS lead too medial.

Also, an important consideration is the angle in the sagittal plane.  For example, a trajectory that is too shallow risks moving obliquely though the short axis of Vim.  This could result in upper contacts in the ventral thalamus pars oralis and only the lowest contact in Vim.  The angle can only be determined by the localization at the top of a 15 to 20 mm trajectory and then at the bottom.
              In targeting the GPi, the homuncular representation related to the most limiting symptoms is targeted.  For example, it does little good to place the DBS lead in the lower extremity region in a patient with cervical dystonia.  Again, the spatial distribution of the homunculus is complex and requires data from a long segment to determine the most optical trajectory.  Also, it is important that a substantial portion of the proper homuncular representation in the GPi be identified.  While detecting head-related units in the last couple of mm’s of a MER trajectory would be important as in the case of cervical dystonia, one cannot know whether the region above the recordings contains sufficient amounts of GPi.  If there is too little GPi, then the trajectory likely is too lateral and DBS ineffective.  Again, this cannot be known without recording over a significant length.

DBS lead testing
               DBS lead test stimulation is an integral component of intraoperative neurophysiological monitoring, regardless of whether MERs are performed.  There are some neurosurgeons, neurologists and physiologists who only test with the stimulation configurations (the combination of active negative and positive contacts on the DBS lead) and stimulation parameters such as pulse width, pulse rate and stimulation current (or voltage) that are commonly reported as effective.  If the patient has a good response in the absence of adverse effects, the DBS lead in permanently implanted.
               This approach presumes that the DBS lead testing in the operating room is sufficiently predictive of what will happen in the clinic postoperatively.  The experience of this author suggests that this often is not the case.  There are many reasons for this.  There may be subtle changes in lead location if there was significant brain shift during the surgery.  Further, there may be subsequent changes in the tissue impedance that would affect response, particularly if constant voltage stimulation is used (constant current stimulation is recommended).
               For these reasons, this author recommends testing with configurations and stimulation parameters that would be worst case.  Typically, these would include monopolar configurations, wide pulse widths (recommend 120 microsec), pulse rate (recommend 185) and high currents (up to 4 milliamps or 5 volts).  The primary issue is that the patient be able to tolerate these worst-case scenarios.

Wednesday, January 11, 2017

Local Field Potentials - What Do They Mean?

Local Field Potentials – What Do They Mean?

Erwin B. Montgomery Jr. MD
The drunkard looking for his wallet that he lost up the road was asked why look beneath the street lamp.  He replied, “The light is better.”
               Local Field Potentials (LFPs) have gotten a great deal of press in the literature related to the pathophysiology of Parkinson’s disease and to Deep Brain Stimulation (DBS).  For example, increased amounts (power) of oscillations in the LFPs in the vicinity of 20 Hz (high beta frequencies) have been thought to cause the motor abnormalities of Parkinson’s disease and have been used as a signal that could control the delivery of DBS in a feedback or closed-loop method.  To be sure, the increased beta power seen in the LFPs of motor control structures of untreated Parkinson’s patients reduces with effective treatment.  However, 15-20% of untreated patients with Parkinson’s disease do not have increased beta power.  Does this mean they have a different form of Parkinson’s disease?  Note, it would be circular reasoning to say that they have a different form just because these patients do not manifest increased beta power.  More likely, increased beta power is not a necessary condition for Parkinsonism.  Further, DBS in the high beta frequencies can improve the bradykinesia of Parkinsonism (Huang H, Watts RL, Montgomery EB Jr., Effects of deep brain stimulation frequency on bradykinesia of Parkinson's disease. Mov Disord. 2014, Feb; 29(2):203-6. doi: 10.1002/mds.25773. PubMed PMID: 24395752.).  Assuming that DBS at these frequencies drives the LFP, high beta power is not a sufficient condition for Parkinsonism.  Things that are correlated, but not necessary or sufficient causes, are typically epiphenomena. 
So what are LFPs and why should increased beta power be a topic of discussion?  One would think this would be the first question asked long ago, but it would seem the horse escaped before the barn door could be closed.  Unfortunately, there is already an extensive body of literature that presumes the significance of increased beta power and, with selective attention and Confirmation bias, any attempts to “walk back” enthusiasm for the notion of increased beta power will not be easy.
A phenomenon that is epiphenomenal is not unimportant or of no use.  For example, the VDRL laboratory test, as an initial screen for syphilis, measures the antibodies to a cardiolipin derived from ox heart muscle tissue, not the Treponema pallidum that actually causes syphilis.  The basis for the VDRL is that a person infected with Treponema pallidum generates antibodies that cross-react with the cardiolipin in the ox heart muscle.  In the case of increased beta oscillations, whatever the real mechanism underlying the pathophysiology of Parkinsonism also produces increased beta power.  However, because the mechanisms are not the same as increased beta power, it may be absent in some patients just as the VDRL may be negative (false) in some patients with syphilis.  Thus, increased beta power in the LFP may be a good biomarker for the presence of mechanisms that produce Parkinsonism, even if not causal to Parkinsonism.  There have been demonstrations of closed-loop DBS triggered off detection of increased beta power but whether this proves to be better than other methods of DBS remain to be seen.
In any event, the use of increased beta power for closed loop control needs to be assessed in the manner of a diagnostic test.  The Positive and Negative Predictive Values need to be determined.  For example, the Positive Predictive Value (PPV) is determined by:

where P(a|b) is the probability of there truly being increased beta power when the detection system indicates the presence of increased beta power, P(a) is the prevalence of true increased beta power in population of patients with Parkinson’s disease, P(b|a) is the probability that the detection system will indicate increased beta power when there truly is increased beta power, and P(b) is the probability of the test indicating increased beta power when applied to all the patients in the population of concern, which includes false positives.  Assuming that 15% of the patients studied do not have increased beta power suggests that the PPV is on the order of 85%.  Whether this probability is sufficient to warrant the use of closed-loop DBS is an ethical consideration based on the cost and risk of false negatives relative to savings related to true positives.  Similarly, analyses can be applied to the determination of Negative Predictive Value.
               The concerns above relate to medical engineering and therapeutic issues.  From a scientific perspective, the important consideration relates to the specificity of b for a which would be related to the probability of detecting increased beta power in those without Parkinsonism, P(b|¬a), where ¬a is not having Parkinsonism.  The specificity would be 1- P(b|¬a).  As P(b|¬a) approaches the value 1, then the less likely it is that increased beta power is a necessary condition for Parkinsonism and thus, increased beta power can be considered to reflect true pathophysiological mechanisms.  To truly assess P(b|¬a) it is necessary to assess P(b) in samples of patients other than those with Parkinsonism.  Ethical concerns limit the range of patients that can undergo testing of LFP beta power in the central nervous system.  However, recordings of LFPs in the subthalamic nucleus of patients with dystonia have a high prevalence of increased beta power (Wang DD, de Hemptinne C, Miocinovic S, Qasim SE, Miller AM, Ostrem JL, Galifianakis NB, San Luciano M, Starr PA. Subthalamic local field potentials in Parkinson's disease and isolated dystonia: An evaluation of potential biomarkers. Neurobiol Dis. 2016 May;89:213-22. doi: 10.1016/j.nbd.2016.02.015. PubMed PMID: 26884091; PubMed Central PMCID: PMC4790450).  These observations suggest that P(b|¬a) would be quite high and supporting the notion that increased beta power is not a sufficient condition for Parkinsonism.
               From the discussions above there are reasons to question both the external and ecological validity of the notion that increased beta power in the LFPs is causal to Parkinsonism.  In addition, there are biophysical reasons to question the face validity of increased beta power directly causal to the pathophysiology of Parkinsonism.  The question is whether LFPs, as customarily used in assessing beta power, actually measure what would be reasonably accepted as neuronal activities relevant to the pathophysiology of Parkinsonism.
               LFPs are typically recorded through electrical contacts (conductors) that are very large relative to the electrical sources generated by individual neurons.  For example, a typical DBS electrical contact is a cylinder that is 1 mm in length and 1.27 mm in diameter for a surface area of 1.99 mm2.  One can contrast that with the electrical contact of a microelectrode typically used in DBS lead implantation surgery for target identification which is on the order of
4.02×10-4 mm2.    The microelectrode recordings are from a volume of tissue on the order of 0.065 mm3, compared to the DBS lead contact volume of 100 mm3.  This means that the microelectrode will include action potentials back propagated into the dendritic trees on the order of 1 to 10 neurons, whereas the LFP will record the summed activities of electrical potentials in the dendritic trees on the order of thousands of neurons.  Typically, the changing electrical potentials of the dendritic trees of the relatively few neurons recorded by a microelectrode can be discriminated based on the time and waveforms of the changing electrical voltages of the action potentials.  No such discrimination is possible with LFP recordings.  Thus, LFPs represent the integrated (weighted sum) of all the changes in electrical voltages in the thousands of dendritic trees in the volume of brain tissue recorded.  There is no way to attribute oscillations in the LFPs to the changes of electrical voltages in individual neurons.  This alone creates doubt that LFPs can provide insight into neuronal pathophysiological mechanisms.
               The situation becomes even more problematic as summing oscillators (time varying neuronal membrane electrical voltages) gives rise to complex interactions such as positive and negative resonance and beat effects (see Montgomery Jr. E. B., Deep Brain Stimulation Programming: Mechanisms, Principles and Practice, 2nd edition, Oxford University Press, 2016).  Beat effects are particularly important when considering LFPs.  When two oscillators interact, another oscillation is created at a frequency roughly equal to the differences of the fundamental frequencies of each oscillator (see figure below).  Two oscillators, at 20 and 40 Hz respectively, will produce a beat frequency of 20 Hz (in the beta range).  However, two oscillators at 120 and 140 Hz, or 220 and 240 Hz, all will produce a beat frequency of 20 Hz.  Thus, any combination of two oscillators whose difference is 20 Hz, will produce a third oscillation at 20 Hz.  The problem is that recording a 20 Hz oscillation in the LFP provides no information as to the nature of the underlying oscillators in the volume of neurons recorded.  This is a manifestation of the Inverse Problem, where any number of mechanisms can produce the same phenomenon and one cannot determine the specific causal mechanism from observation of the phenomenon alone.  The potential that LFPs are recordings of beat interactions further casts doubt on the face validity of LFPs reflecting underlying neuronal pathophysiology and physiology.

Figure. Schematic representation of the interactions between two oscillators A and B.  The sum of A and B produces D in which there is a raise and fall of the amplitudes.  This rise and fall occurs over a slower frequency and is called the beat frequency.  Generally, the beat frequency is proportional to the difference in the frequencies of the two oscillators.  However, if the beat interaction passes through a low pass filter where the cutoff frequency is considerable less than the frequencies associated with A and B, then the faster components (higher frequency components) of D will be filtered out, resulting in a low frequency oscillator C.
               The situation becomes even more problematic.  It is quite possible that some of the neuronal oscillators may be relevant to the neuronal pathophysiological mechanisms, while others may be irrelevant.  The LFP cannot discriminate between those oscillators that are relevant and those that are not.  The LFP will be the result of the beat interactions between relevant and irrelevant oscillators.  There is no way the investigator can be sure that irrelevant neurons are not being recorded.
               From the figure above, one could argue that if the increase in the beta frequency LFP is the result of interactions between two oscillators with very high frequencies, certainly some element of the high frequencies should be seen.  However, the typical filter settings used in LFP recordings are such that high frequencies are filtered out.  When care is taken to have high cutoffs in the low pass filters, oscillations of high frequencies are found.  LFPs, in a great many studies, show that the power spectrum (the “amount” of oscillations at specific frequencies) follows the power law of  (Milstein J, Mormann F, Fried I, Koch C (2009) Neuronal Shot Noise and Brownian 1/f2 Behavior in the Local Field Potential. PLoS ONE 4(2): e4338. doi:10.1371/journal.pone.0004338). Thus, it is likely that neuronal oscillators in the LFPs would be hard to detect, particularly if the power spectral density of the LFP is normalized to the total power.
               Certainly, there are methods to try to minimize some of these complications.  For example, evoked potentials constructed from LFPs to specific stimuli can help lessen the impact of oscillations from irrelevant neurons that cannot be excluded from the LFP alone.  The presumption is that the change in the dendritic neuronal membrane electrical voltages are time (phase) locked to the stimulus, whereas the irrelevant neuronal oscillators are random and tend to cancel out.  Thus, LFP evoked potentials are very useful.  However, the assumption that these evoked potentials reflect pathophysiology and physiology at the neuronal level is best avoided or at least tempered by appending appropriate caveats.
               It is also true that LFPs are technically much easier to obtain.  Lower impedance electrodes help to reduce electrical artifact.  In the case of DBS, the choice of a therapeutic DBS lead constrains the choice of recording electrodes, particularly over longer recording times (indeed it is possible to implant pulse generators that record LFPs from the DBS leads indefinitely).  However, the convenience of LFPs should not be confused for credibility.
               While it might well be that increased beta oscillations in the basal ganglia-thalamic-cortical system represent some form of idling in which the patient’s brain gets stuck and the patient cannot move, demonstration of increased beta power in the LFPs is not evidence.  Further, demonstration of increased beta power in the LFP is also consistent with any number of alternative hypotheses.  Failure to consider these alternatives commits the Fallacy of Limited Alternatives because of Confirmation Bias.  The Fallacy of Limited Alternative occurs when a limited set of alternative explanations is considered.  The denial of some alternatives is thought to increase the credibility of the alternatives that remain.  In probability, this results in a form of the Gambler’s Fallacy, in which reducing the probability of one alternative is thought to increase the probability of the other.  For example, being dealt a poor hand in previous rounds of poker leads the card player to think the next round has a higher probability of yielding a good hand.

Monday, November 28, 2016

Rechargeable Implanted Pulse Generators & Directional Leads for DBS: An Exercise in Ethics

 November 28, 2016

Erwin B. Montgomery Jr. MD
Volume 2, Number 10

Resident – Regarding the use of non-rechargeable implanted pulse generators and directionally steerable leads, isn’t the situation one of “penny wise, pound (English equivalent of the dollar) foolish”?

 Attending – That depends on who is spending the penny and who is spending the pound.

           Clinicians now have choices in the type of implanted pulse generator (IPG) and directionally specific leads to use in providing patients Deep Brain Stimulation (DBS).  For example, the clinician may implant the rechargeable or a non-rechargeable IPG.  Clinicians can chose  DBS leads with electrode contacts that are segmented, or in pieces, around the lead.  Previous leads had contacts that were continuous around the circumference.  The advantages and disadvantages are reviewed here.    Whether a rechargeable IPB or directional lead is ultimately used, perhaps the more important part of the selection process is how the decision was made.

Shaping and positioning the volume of tissue activation relative to the regional physiological anatomy surrounding the DBS lead’s electrical contacts is critical to successful DBS therapy.  Leads with continuous circumferential contacts can allow for positioning in the long axis of the lead but not in the plane orthogonal to the long axis of the lead.  For example, if the DBS lead in the subthalamic nucleus is too anterior, and therefore is close and parallel to the posterior limb of the internal capsule, positioning the volume of tissue activation in the long axis may not avoid tonic contractions at stimulation currents necessary for symptomatic relief.  For these patients, there may be no other choice but complete surgical replacement of the DBS lead.

With leads whose electrical contacts are segments in the plane orthogonal to the long axis of the DBS lead, it may be possible to stimulate through the posterior projecting segment, thereby avoiding stimulation of the posterior limb of the internal capsule.  This may allow the patient to benefit and avoid a second DBS lead implantation surgery.  However, it is also possible that the patient receiving the DBS lead with continuous circumferential contacts may settle for compromised benefit, a phenomenon this author describes as the tyranny of partial benefit.

The situation of having to undergo a surgical revision of the DBS lead or to compromise on benefit may not occur very frequently.  In advanced centers, the need for DBS lead surgical revision may be on the order of 2%.  At one institution, it was 10%.  Unfortunately, these percentages do not include those who elect compromised benefit.  Even if these occur 10% of the time, the problem is there is no way to predict which individual patient out of 10 will have compromised benefit that could have been prevented by placement of a directional DBS lead.  If the clinician decides that it is of sufficient value to do all that is reasonable to avoid the potential problem, he or she may elect to place the directional DBS lead in every patient.

               One significant advantage of the rechargeable IPG is the reduction in the number of surgeries required to replace expended IPGs.  The reduction in the number of surgeries significantly reduces the risk of infections that could require explantation of the entire DBS system, failure of therapy, and surgery to implant a completely new DBS lead.  The increased risk of infection is not simply additive.  Each surgery increases the risk for the subsequent surgery because of the accumulation of relatively avascular scar tissue.

               Another significant advantage relates to postoperative DBS programming.  Initially, much of the strategy underlying DBS programming was directed first at maintaining battery life with non-rechargeable IPGs.  Minimum pulse widths, frequencies and stimulation currents (voltages) were recommended (Montgomery Jr., E. B., Deep Brain Stimulation Programming: Principles and Practice, Oxford University Press, 2010).  Those recommendations remain true when non-rechargeable IPGs are used, but they may no longer hold true with the advent of rechargeable IPGs.

               In either case, the endpoint was giving the patient satisfactory control of symptoms, signs and disabilities.   Whatever currents (voltages), pulse widths, frequencies, and electrode configurations necessary for sufficient control dictated programming.  However, starting at minimal stimulation parameters and electrode configurations often would require considerable and extended programming efforts.

               With the use of rechargeable systems, minimizing current drain from the IPG battery is no longer the first concern.  Thus, decisions about pulse width, for example, can be driven by what is clinically optimal rather than what is minimal in terms of battery depletion.  There is considerable evidence that starting at a pulse width of 120 microseconds may be the most effective when compared to the typical starting pulse width of 60 or 90 microseconds (Montgomery Jr., E. B., Deep Brain Stimulation Programming: Mechanisms, Principles and Practice, 2nd edition, Oxford University Press, 2016).  This alternative approach, which is based on what is electrophysiologically optimal rather than what is electronically optimal,  may greatly simplify post-operative DBS programming.  This could translate into greater efficiency, which translates to earlier and better symptomatic control and less expense.

               There are disadvantages to consider, the first of which is the risk of IPG failure due to lack of adequate recharging.  Certainly, there are case reports of catastrophic exacerbations of symptoms, signs and disabilities of the disease being treated.  Further, there have been reports of a neuroleptic-malignant-like syndrome in patients with Parkinson’s disease following IPG failure.  However, this does not appear to be a common problem with the accumulating experience gained from using rechargeable systems.

               The second disadvantage is related to the greater costs of the rechargeable IPG, although it is not clear what drives the increased costs.  It is this increase in costs, and who ultimately bears the burden of those costs, that is the source of ethical difficulties.

The ethics of the rechargeable versus non-rechargeable IPD and directional versus non-directional DBS leads

               If there were no financial cost issues, it is likely that there would be little opposition to the preferred use of rechargeable IPGs or directional DBS leads.  That patients are not being offered rechargeable IPGs or directional DBS leads suggests other factors are at play, and these are not solely medical or what is in the best interest of the patient.  This is not to suggest that considerations other than the best interest of the patient are not relevant, but rather the issue is how these other interests  are leveraged in the decision-making process.  Thus, the question is one of ethics.

               A strategy in resolving ethical questions is to identify the shareholders (those that have the power to shape the decision) and stakeholders (those that are affected by the decision).  Clearly, the surgeon implanting the IPG is a shareholder.  The person paying for the IPGs is a shareholder; typically this is a governmental organization or a commercial insurer.  Indirectly, IPG and DBS lead manufacturers may be shareholders to the extent that they influence cost considerations and market to surgeons, governmental organizations and commercial insurers.

               It is not exactly clear how the patient or the patient’s representative acts as a shareholder.  In the negative sense, the patient or representative could deny the use of the IPG or DBS lead offered.  To use the IPG or DBS lead against the consent of the patient or representative could constitute the criminal act of battery.  However, it is not clear how the patient can compel the use of one IPG or DBS lead over others.

               Even with the lack of a patient’s or representative’s power to compel the use of a specific IPG or DBS lead, there remains the requirement of the surgeon to obtain informed consent.  This means that, at a minimum, the patient or representative must be informed of all the reasonable alternatives and the implications of the choice.  Note, it is not what the surgeon considers reasonable; rather it is the information a reasonable patient or representative would want to consider in their decision-making.  The advantages and disadvantages described above would be relevant to obtaining informed consent.

               What latitude the surgeon should allow the patient or representative in the IPG and DBS lead decision-making is unclear.  Typically, the answer  is linked to what is considered to be the standard of practice.  Do similarly-situated, reasonable surgeons offer the patient or representative the choice?  If so, then every surgeon is bound to offer the patient or representative the same choice.

               Surgeons and the patient community (meaning the patient, family members, caregivers and others affected by the health of the patient) are stakeholders.  Also, society at large may be a stakeholder.  There is another very important stakeholder, those responsible for the postoperative management of the patient.  It is critical for all to understand that the benefits of DBS, particularly as these outweigh the risks and costs, only accrue with actual brain stimulation.  Up to that point, including surgery, it is all risk and cost.

               Stakeholders who are not direct shareholders, such as neurologists providing the postoperative management, can become shareholders to the degree that direct shareholders, such as surgeons, cede some shareholder status.  For example, surgeons could acquiesce to neurologists responsible for the postoperative management.  It is not clear that this is the usual practice or that neurologists would be expected to share in the IPG or DBS lead decision-making.

Balancing shareholder’s and stakeholder’s interests

               The relations between all the shareholders and stakeholders are complex and almost always involve some conflict of interest.  For example, whoever “pays the bill” will be in conflict with the stakeholders if the reasonable choices involve differences in costs.  How then does one resolve the conflicts of interest?

               Ethical principles relate to how people treat each other.  At times, these ethical principles are codified in law or legal case precedents.  More commonly, and for a variety of reasons, conflicts between shareholders and stakeholders are “settled out of court.”  In order to apply ethical principles, it is helpful to examine the pair-wise relationships among and between the shareholder and stakeholders.  Once all the relationships are identified, one examines how these relationships relate to the specific ethical principles.  One set of ethical principles includes autonomy, beneficence, justice, and non-malfeasance.  The last term, non-malfeasance, means not causing harm rather than making a medical mistake. 

               Autonomy, in its most general sense, means respect for the patient as the ultimate stakeholder.  This does not mean that the patient has the final decision.  As noted, the only absolute final decision available to the patent or representative is to forego the medical care offered.  However, the patient or representative must be given a full and complete understanding of what the patient or representative is being asked to decide.  But autonomy is not unidirectional.  The surgeon as an agent may have autonomy, though not in the sense that the surgeon has unconditional decision-making authority.  One notion of autonomy is taken from the work of Immanuel Kant, in which each person is an end unto him or herself first and not a means to someone else’s end.

               Beneficence is the motivation to do good.  In biomedical ethics, it is not necessarily an obligation to do good in that the failure to do good is a failure of the ethical principle.  Typically, one is not obligated to enter a burning house to rescue someone.  There may be some obligation if the person is a firefighter.

               Non-malfeasance is not to do harm.  However, even this is conditional.  In the course of providing beneficence through DBS lead implantation, the surgeon will create harm if for no other reason than making an incision in the scalp.  Thomas Aquinas in his Summa Theologica (1485 C.E.) provided some guidelines to balancing beneficence and non-malfeasance in his Principle of Double Effect.

               Justice is perhaps the most difficult principle to define.  What justice is relates to the prevailing moral theory.  For example, in Utilitarianism, what is just is that which has the greatest benefit, however that is defined.  Most cases of ethical conflict arise when consideration of the principles of beneficence, autonomy, and non-malfeasance fail to create consilience among all the shareholders and stakeholders.  In these situations, the consideration of justice, interpreted in light of the prevailing moral theory, is used to adjudicate the result.

               The patient or representative.  On the axis of beneficence, nearly all shareholders want good for the patient.  The sticky question is what is the obligation of shareholders to beneficence.   This can be considered in terms of cost (malfeasance or harm) to the shareholder.  In the absence of any cost, shareholders are expected to provide beneficence.  When there is a cost to the shareholder, then other considerations are required.

               Typically, the obligation to beneficence on the part of the governmental organization or commercial insurer to the patient is contractual, thus invoking a Libertarian theory of justice.  The Libertarian moral theory holds the maximum good is the greatest freedom.  What prevents anarchy is the willingness to give up some freedom in exchange for some beneficence, for example a functioning society.  In the case of a commercial insurer, the freedom exchanged is obligation to pay for the beneficence given to the patient in return for something of value, such as wealth in the form of a premium.  The contract that exits between the insurer and the patient is what has been mutually agreed to and includes societal laws which mediate the contracts.  Thus, whether or not a rechargeable or non-rechargeable IPG or a directional DBS lead is implanted may be stipulated in the contract, more or less.

               Based on a Libertarian moral theory, the obligation to beneficence on the part of surgeon to the patient is less clear.  The patient or representative lacks knowledge of sufficient detail to be expected to negotiate a contractual agreement to bind the surgeon.  It is the standard of practice in the profession and governmental laws and policies that define the surgeon’s obligations to beneficence.   Within those constraints, the patient has every right to expect beneficence and at the very least to be fully informed of all reasonable alternatives, such a rechargeable IPGs and directional DBS leads.  The problem arises when the surgeon becomes an agent for the governmental organization or commercial insurer and limits or conditions the expected beneficence; unless of course, the patient or representative are in full understanding that the surgeon also is acting as an agent for those other than the patient.  On the axis of autonomy, the patient or representative has the right to be fully informed by all shareholders and surgeons, as well as those paying the bill.

               A significant difficulty ensues where explicit laws and contract terms are not available.  Typically, such contracts are not so fine grained.  Often, analogous situations can be sought to gain some notion of precedence, legal or otherwise.

               Other moral theories will define justice differently.  For example, an Egalitarian moral theory would suggest that either all or no patients could expect a rechargeable IPG or a directional DBS lead.

               The surgeon.  The notion of autonomy extended to the surgeon is complicated.   Again, borrowing from Kant, the surgeon is an end in his or herself and not as a means to others.  The means that the surgeon is subjected to cannot outweigh the surgeon’s own ends.  In other words, the surgeon cannot be expected to sacrifice his or herself where the costs to the surgeon exceed the benefit in return.  In other words, the surgeon cannot be expected to assume harm or malfeasance.

               The surgeon’s obligation to benefit the patient is complex.  In part, the obligation is contractual.  The surgeon is granted a privilege that others are not, analogous to a monopoly.  Not just anyone can practice surgery.  That privilege is granted in exchange for an obligation to patients that are citizens to the state or province that grants such privilege.

               Another implicit contractual obligation is to the citizens who subsidized the surgeon’s education, allowing the surgeon the privilege of being a doctor.  It is highly unlikely that any physician bore the full cost of his or her education.  Otherwise, becoming a physician is like a lottery prize and a matter of luck.  Just because an applicant to medical school may have the highest grades, there is no obligation for the citizens to award that person what amounts to a lottery prize free of obligation.

               A problem arises when the surgeon assumes the limits to obligations from governmental organizations or commercial insurers as an agent prior to and in the absence of consent by the patient or representative.  In other words, the surgeon cannot be the agent for the government or the insurer without prior consent of the patient or representative.  Thus, the surgeon has an obligation to at least offer rechargeable IPGs or directional DBS leads, though the surgeon is not expected to bear the financial consequences.  It is up to the patient or representative to make a decision based on the patient’s situation.

               The neurologists and those providing postoperative management.  As stated, the only benefit that is accrued is when the stimulation has been initiated and optimized.  Thus, the surgeon’s responsibility extends beyond the implantation of the DBS systems; he or she must ensure appropriate postoperative management.  The neurologist and others providing the postoperative management are extensions of the surgeon’s obligations to beneficence.  Hence, surgeons have an ethical relation to the neurologist and those providing postoperative management.  These obligations are to beneficence, autonomy, and justice, to the neurologist and others providing postoperative management.

               Those providing the postoperative care should be provided the means to do good.  At the first analysis, the question becomes who is best to decide which DBS systems, rechargeable IPGs and directional leads, is most optimal for the patient.  Unless the surgeon is also an expert in postoperative DBS management, then it is the neurologist and those providing the postoperative management who are in the better position to judge.  Hence, the surgeon has the obligation to autonomy that is respecting the judgment of those providing postoperative care.  The best way to fulfill that obligation to autonomy is to preoperatively engage the neurologist and those providing the postoperative management in the decision-making process.

               The governmental organization.  Typically, the ethical obligations related to governmental organizations center on the concept of sovereignty as codified in laws and regulations.  However, given the complexity of the human condition, particularly in disease, it is impossible to anticipate every eventuality in a fine-grained manner.  Further, laws that are too vague in order to be all-encompassing generally fail legal challenge.  Laws can be expanded in range by appending concepts such as standards of practice, appeals to what the reasonable person might do, and case precedence that provides context.  Certainly, all other shareholders and stakeholders have obligations of autonomy, non-malfeasance and justice to governmental organizations.  It is not clear whether this extends to an obligation of beneficence to the government that goes beyond the more contract-like obligations to laws and regulations.  Further, one would always hope that the citizens could change the laws and policies through their representatives in government.

               Commercial insurers. This term is used in the context of those paying the bill but excludes governmental organizations.  In the United States, private insurance companies, rather than Medicare or Medicaid, would be the subject.  For commercial insurers, issues of obligation center on contracts.  Typically, obligations of beneficence to patients are limited to contract specifics and law, except certain mandates resulting from the Affordable Care Act of 2010.  Notions of obligations to the standards of patient care typical of physicians and healthcare professionals do not figure in relations with commercial insurers.  Indeed, the Employee Retirement Income Security Act of 1974 (ERISA) have written this distinction into law.

               For-profit commercial insurers have an obligation of beneficence to their shareholders that clearly risks a conflict of interest with respect to the insurers’ obligations of beneficence to the patients whose healthcare the insurers pay.  How these conflicts in the obligations to beneficence are resolved is unclear.  In societies where the majority of care is covered by commercial insurers, there is a need for stable and successful insurers.  The question that remains is what is the price of stability extracted from patients and the rest of society.

               There is a presumption among laissez-faire capitalists that in an open and free market, the “invisible hand” of the market would optimize the relationships between insurers and insured.  However, the healthcare market is anything but open and free, and thus the assurances of laissez-faire capitalism are suspect.  Generally, patients or representatives are not free agents, as they cannot exercise independent judgment based on sufficient understanding.  Depending on anyone else, the patient becomes married to the advisor’s interests and obligations.

               At first, it would seem that the contractual nature of the relationships between commercial insurers and the patient, surgeon, neurologist and those providing postoperative management would be straightforward, but this would be naive.  In the 1980s, insurers wielded threats of non-renewal of contracts with healthcare providers to enforce obligations to the insurers beyond those stipulated in contracts, such as gagging physicians from educating patients and representatives about options based solely on medical concerns.

               Society.   Clearly, society is a stakeholder in the medical decisions made.  Ultimately, society will pay in one way or another if a rechargeable IPG or a directional DBS lead is not offered.  There will be harm done to the national “pocketbook,” as well as the nation’s reputation, even if the scale of such loss is minuscule by national standards.  But if society holds that healthcare is a right, by whatever justification, analogous to a person’s civil rights, then just as it is hard to accept that only 99% of citizens have their civil rights respected, it would be hard to accept patients not be offered rechargeable IPGs and directional DBS leads.  To think otherwise is to say that some person’s rights are expendable.

               The preamble to the Declaration of Independence of the United States holds that a person’s right to life, liberty and the pursuit of happiness is based on Natural Law.  However, the arguably does not to have the force of law.  Rather, the practical operations of government seem to hinge on enlightened self-interest.  Shareholders may force actions on stakeholders when the shareholders see themselves at risk.  In healthcare, there are quarantines, reporting of communicable diseases and gunshot wounds, and forced vaccinations, among other measures.  The notion that healthcare is a right that must be supported by all citizens reflects a deontological or Kantian moral theory.  The alternative that everyone has a right to enter into a contract to obtain healthcare, but not necessarily to be guaranteed healthcare, is in line with a Libertarian moral theory.

Medical care has advanced to a point were risks can be increasingly compartmentalized.  Diseases among certain stakeholders are no longer seen as a risk to the shareholders.  Tuberculosis in a stakeholder may readily constitute a risk to the shareholder and thus invoke public health law measures.  It may well be that a stakeholder’s Parkinson’s disease is not seen as a credible threat to a shareholder.   The risk of Parkinson’s disease for any shareholder is likely small and the means to mitigate any effect should the shareholder develop Parkinson’s disease may not require optimal DBS treatment utilizing rechargeable IPGs and directional DBS leads.  The new Golden Rule may have become, do onto others what you would want done for you, unless you can avoid it.

 Risk management

Support of beneficence to patients in the context of health insurance has always been an issue of risk management.  In the Egalitarian moral theory, risk management is the amortization of risk over the entire citizenry.  The healthy pay more in comparison to what is spent on them as a way of paying forward when the healthy later become ill or disabled.  However, this is difficult.  Ask any young healthy person whether they would want a rechargeable IPG or directional DBS lead in the possible (but unlikely) future, they may say yes in the abstract; but test their willingness to pay for even a part of it now.  It is not surprising that the “individual mandate” was seen as critical to the success of the Affordable Care Act and how any insurance program that does not include some aspect of the “individual mandate” is not likely to be successful.

               Another approach to risk management is risk avoidance.  There are as many types of risk avoidance as there are obligations to beneficence.  For example, an insurer, either governmental or commercial, has an obligation to beneficence to the patient.  However, constraining the beneficence can mitigate the risk entailed.  This can be seen by the exclusion of coverage for pre-existing medical conditions, caps on benefits, and panels of providers that are insufficient or impractical to meet the patient’s needs.  A simple way to avoid having to provide the beneficence of a rechargeable IPG or directional DBS lead is simply to not have a surgeon in the network of providers willing or able to provide DBS at all.  Managing the beneficence to the surgeon in order to avoid risk can be accomplished by capitation or reimbursements that effectively make provision of a rechargeable IPG or directional DBS lead too costly to the surgeon.

 A final answer?

               The purpose here is not to demand a specific answer to whether a rechargeable IPG or directional DBS lead should be offered, though the author’s sympathies should be fairly apparent.  In biomedical ethics, context is all-important; and thus, flexibility in the answers likely is necessary.  However, flexibility is not license for any particular shareholder to do as they please.  Rather, whatever decision is reached in any particular situation should be based on a full and disciplined assessment of all the concerns of all those involved.


Monday, October 3, 2016

Greenville Neuromodulation Center Professional Newsletter
Volume 2, No. 9
September 30, 2016
Caution Advised - Deep Brain Stimulation (DBS) is Not Normal
and Does Not Restore Normality
Erwin B. Montgomery Jr. MD
Professor –     As Abraham Maslow said, “when all you have is a hammer, the whole world looks like a nail.”

Student –       The statement presupposes that one knows that something other than a hammer exists even if it is not available.  It would not make sense if only “hammers” existed because there would be nothing to differentiate “hammer” from anything else.  Indeed, there would be no word “hammer.”  The saying must imply that there are known alternatives to hammers, yet still the whole world looks like a nail.  What is that?

Professor –      Willful ignorance.

Student –         I would call that being stupid.

Professor –      I was trying to be kind.

Student –         They need to “own it.”

               In addition to the remarkable clinical benefits attributable to Deep Brain Stimulation (DBS), it has become an increasingly used tool to investigate the operations of the brain.  Scientists examine the effects of DBS on a specific target, such as the subthalamic nucleus, for various physiological or psychological functions, to infer the role of the target in the function.  In the large majority of research publications, the effects of DBS are attributed to the structure being stimulated. Further, the effects are then thought to reflect the physiology or pathophysiology of the structure targeted by DBS.

               Such inferences are invalid.  This is not to say the inferences are true or false, only that the experiments (arguments) that contain the inference are invalid.  For example, consider an experiment examining the effects of DBS targeted at the subthalamic nucleus on lexical function.  Differences are noted in tests of lexical function when the stimulator is turned on compared to the situation when the stimulator is turned off.  The inference is made that the subthalamic nucleus is involved in lexical functions.  This is an example of the logical fallacy Post Hoc Ergo Propter Hoc (since b follows a, a must be the cause of b).  This fallacy is a special case of the Fallacy of Limited Alternatives because b could be due to any number of causes other than a, particularly if a is complex and multifaceted.

In the case above, DBS in the vicinity of the subthalamic nucleus generates action potentials in virtually any axon in the vicinity. These include axons terminating on neurons within the subthalamic nucleus, which affects the next or post-synaptic neuron but also antidromically activates the neurons outside of the subthalamic nucleus.  If these neurons have axon collaterals to other structures, then antidromic action potentials will proceed down the branch in an orthodromic manner to activate structures that have nothing to do with the subthalamic nucleus.  Indeed, axons passing in the vicinity without any connection to the subthalamic nucleus may be activated.

The situation becomes even more complex.  Within tens of milliseconds, activities generated in the vicinity of the DBS lead, such as in the subthalamic nucleus, can spread extensively.  This spread is reinforced by resonance with repeated DBS pulses.  Such extension with repeated pulses has been demonstrated by wider distribution of the electroencephalographic evoked potential to a brief train of DBS pulses in the vicinity of the subthalamic nucleus compared to single pulses (Baker KB, et al. 2002).

The DBS pulse is fairly indiscriminate in which axons within the volume of the electrical field are activated. Factors include proximity to the cathode, myelination, axonal diameter and axonal branching patterns.  The DBS pulse stimulates anything where it is put, and targeting seldom obeys physiological organization.  Further, the action potentials initiated in the axons are simultaneously affecting a great many axons.  The DBS pulse is more analogous to a lightning bolt striking a large group of people and then inferring that the mass convulsions are a normal phenomenal.  It is not as though a single person is struck and the effects on those remaining can be understood based on the physiological properties and contributions of that single strike.

None of the criticisms offered above is to say that DBS is not a valuable tool in research, quite the contrary.  Clearly the clinical effects of DBS demonstrate the interactions between the DBS pulses and the physiology.  Rather, the criticism is necessary to properly use DBS as a tool, particularly in drawing inferences from DBS experiments.  Given the complexity of the responses to DBS, including those without any direct relationship to the target of DBS, how can anybody reasonably ascribe the effects of DBS to the physiology of the target?  Specifically, how can anybody reasonable ascribe the effects of DBS in the vicinity of the subthalamic nucleus to physiology of the subthalamic nucleus?  Note, anyone can say just about anything they want to say.  However, if one wanted to be scientifically responsible, then one would withhold attribution or at the very least append the appropriate caveats.  Yet this is seldom done.

One might argue that scientific reports making inferences attributing some property or function to the DBS target really is only describing the “facts” of the experiments and that inferences are only secondary or peripheral.  They may argue that the reader is free to draw whatever inferences she may want.  However, such as claim is not only disingenuous and self-serving, it is a disservice to science.  It reduces science to “stamp collecting”.  If that position was seriously adopted, then science would be reduced to the Solipsism of the Present Moment where no knowledge exists beyond very narrow and exact circumstances of a particular experiment.  There would not be any generalizable knowledge.  Most importantly, the actual observation resulting from a particular experiment could not be used to justify or support any future experiment.  Science always would be just a set of dots (each dot corresponding to a specific and limited observation) and there would never be any lines connecting the dots.  These lines are critical because it is interpolation between the dots onto the line and extrapolation of the lines connecting the dots that point to possible future knowledge.

Inferences are very important and consequently, it is important that inferences are sound.  The epistemic importance of inferences is demonstrated above.  The practical importance of inferences is seen in scientific publications.  Most scientific publications are structured with introductions and discussions which serve the important purpose of connecting previous dots (in the introduction) to justify and set the context for the new dots (experimental observations).  Finally, the new and old dots are interconnected by lines of inference to provide new knowledge and understanding and very importantly point the way to future research.  Yet, it the great majority of scientific publications, the old dots are the past inferences and rarely if ever the actual observations of previous experiments.  Thus, the past inferences take on the epistemic status of facts as they are used in lieu of the original observations in the argument that comprises the research.  Inferences, regardless of their lack of soundness, become quasi-facts.

Most inferences derive from the hypotheses used in the scientific method.  A hypothesis not refuted during the application of the Scientific Method is then taken as true and becomes the inference derived from the observational data.  However, there is a flaw.  The Scientific Method where a hypothesis gives rise to a testable prediction that if found is thought to validate the hypothesis.  However, this use of the Scientific Method is the Fallacy of Confirming the Consequence which is of the form if a (hypothesis) implies b (prediction) is true and b is true, then a is true.  However, b could be true for many reasons other than a.  In the case of DBS in the vicinity of the subthalamic nucleus inducing a change in behavior or functions, the Scientific Method becomes if the subthalamic nucleus is involved in behavior A or function B, then DBS of the subthalamic nucleus should affect behavior A or function B, the experiments demonstrate this to be the case, thus the subthalamic nucleus is involved in behavior A or function B.  However, as discussed above, DBS affects far more than the neurons in the subthalamic nucleus and these other effects could be responsible for the change in behavior A or function B.  There is no way from the experiment alone to know which of the many possible responses to the DBS pulse is responsible for the change. Note, this does not mean that an effect of the DBS pulse directly on subthalamic nucleus neurons cannot mediate the change in behavior or function; only that one cannot conclude that is the case from the experiment alone.

Even if one were to grant that a change, such as a restoration of normal function, was consequent to DBS, this is not to say that the manner by which DBS restores the appearance of normal function is the same as the normal intact physiological mechanisms.  The inability to attribute the mechanisms by which DBS restores a normal function to the same mechanisms that operate normally is due to the Inverse Problem.  Typically, behaviors are mediated by the actions of the lower motor neurons in the brainstem and spinal cord.  These neurons receive multiple and diverse inputs.  Consequently, one cannot know from observing the consequence of lower motor neuron activations what is the source of the relevant actions onto the lower motor neurons.

To argue that normalization of behavior with DBS utilizes the same physiological mechanism normally involved in the behavior would be the logical Fallacy of Four Terms.  Therefore, one cannot attribute presumed DBS-related mechanisms as providing insight into normal physiological mechanisms.  The fallacy would be the following: Major premise -  Normal function implies normal physiology; Minor premise - DBS restores normal function; and Conclusion – DBS is normal physiology.  It would appear there are two terms, “normal physiology” and “DBS” that are linked through the third bridging term “normal function.” However, as described above, the “normal function” in the case of the normal condition may not be the same as “normal function” in the case of DBS.  Thus, there are two different bridging terms, hence the fallacy of Four Terms.  Again, it is important to appreciate that the fallacy does not mean that the DBS mechanisms cannot be the same as the normal mechanisms, but only that the experiment cannot provide confidence that they are the same.  Once again, it is not that the scientists cannot offer the theory that normal mechanisms underlying the normal function is similar to DBS mechanisms.  However, such inferences can only be tentative and thus, demand that the appropriate caveats be explicitly appended to any inferences, which is seldom done.

As can be appreciated, the inferences derived from demonstrating the predictions of hypotheses using the Scientific Method are indeterminate except when the experiments fail to demonstrate the prediction.  In the latter case, failure to demonstrate the prediction is clear evidence for the falsehood of the hypothesis.  Also, the Inverse Problem demonstrates that the same phenomena may have many causes and the phenomena alone cannot determine which cause is relevant.  One has to bring knowledge extraneous to and independent of the experiment to help adjudicate which of the multiple causes is most probable.  But it is important to note, the additional evidence cannot be directed solely at the inference that is being made.  The additional evidence must be directed at all the reasonable alternative inferences.  Failure to do so represents Confirmation Bias.  Yet, rarely in scientific reports are alternative inferences and the relevant external knowledge ever raised.

The tendency towards limited reviews within the introductions and discussions of scientific papers risking Confirmation Bias has a long history.  Charles Bazerman analyzed the rhetoric of scientific reports in the Philosophical Transactions of the Royal Society from 1665 – 1800 (Bazerman, C. 1997, pages 169-186).  In the essay, Bazerman described four stages with indistinct boundaries.  In the first period, 1665-1700, scientific papers were uncontested reports of events.  In the second stage, from 1700-1760, experimental results appeared and the discussions centered over the results.  It was not until the third period, 1760-1780, that papers “explored the meaning of unusual events through discovery accounts (Bazerman, C. 1997, page 184), which can be taken as discussions of the inferences drawn from the accounts. 

During the fourth period, approximately 1790-1800, experiments were reported as claims for which the experiments were to constitute proof, but most were the Fallacy of Confirming the Consequence as described above.  Further, experimenters were presenting claims that were solely the result of their individual efforts and not “recognizing the communal project of constructing a world of claims… Although the individual scientist has an interest in convincing readers of a particular set of claims, he does not yet explicitly acknowledge the exact placement of the claims in the larger framework of claims representing the shared knowledge of the discipline” (Bazerman, C., 1997, page 184).

The operative words are “communal project” and “shared knowledge of the discipline.”  Since the founding of the Royal Society of London in 1661, the community charged with the scientific project was highly selective (Shapin S., Schaffer S., 1985).  Public demonstrations of experiments meant to vouchsafe the results were limited to a select group, perhaps a very early instantiation of Confirmation Bias.  Critics, such as the famous natural philosopher, Thomas Hobbes, were vigorously excluded.  Certainly, the historical analysis of the progress of modern science by Thomas Kuhn (1962) demonstrates that this sort of exclusivity maintains Confirmation Bias even today.

               In the final analysis, the state of knowledge regarding DBS is such that inferring neurophysiological functions from DBS effects is not possible.  Rather, only possibilities can be offered and consequently, all reasonable possibilities should be considered.  Papers attempting to do so have the obligation to clearly state the tentative and problematic nature of such inferences, which is best accomplished by vigorously presenting alternatives based on a thorough understanding of all the relevant literature.  Editors of scientific journals should insist that this be done.


Baker KB, Montgomery EB Jr, Rezai AR, Burgess R, Lüders HO. Subthalamic nucleus deep brain stimulus evoked potentials: physiological and therapeutic implications. Mov Disord. 2002 Sep;17(5):969-83. PubMed PMID: 12360546

Bazerman, C. “Reporting the Experiment: The Changing Account of Scientific Doings of the Philosophical Transactions of the Royal Society, 1665-1800”, in Landmark Essays on Rhetoric of Science: Case Studies, Vol. 11, R. A. Harris ed., Hermagoras Press, 1997, pages 169-186

Kuhn T. The Structure of Scientific Revolutions, University of Chicago Press, 1962

Shapin S, Schaffer S. Leviathan and the Air-Pump: Hobbes, Boyle, and the Experimental Life. Princeton University Press, Princeton, New Jersey, 1985, ISBN 0-691-08393-2