The Effects of Neuroleptics on Plasma Homovanillic Acid

Arnold J. Friedhoff and Raul R. Silva

One of the most dramatic observations in psychopharmacology is that various changes in plasma homovanillic acid (pHVA) are highly correlated with the therapeutic response to neuroleptic drugs. In an early study, Bowers' group (4) showed that patients with schizophrenia who responded to a neuroleptic had a decline in pHVA over several weeks of treatment, whereas nonresponders showed little change from baseline. Subsequently Pickar et al. (19) found an impressively high correlation (0.82) between improvement in symptoms of schizophrenia and decline in pHVA. More recently Davila and colleagues (11) found that the extent of the increase in pHVA that occurs early in treatment (4 days after initiation of treatment in this study) was also predictive of therapeutic outcome. A vigorous increase in initial response was predictive of a good therapeutic outcome as was a later decline. The ability to predict outcome after only a few days of treatment could be useful in making decisions about whether to continue treatment with a given drug.

Davis et al. (12), looking at pHVA from a different perspective, reported that the severity of symptoms as schizophrenia and basal pHVA levels were correlated. In the interim, many other investigations have confirmed all or part of these early observations. Green et al. (15) confirmed the previous findings of Davila et al. (11) showing that an early increase in pHVA followed by a subsequent decrease in pHVA, was a good predictor of a favorable clinical response. Chang et al. (6) also found pretreatment pHVA levels higher in patients who turned out to be good responders to haloperidol when compared with a poor responder group. The good outcome group also showed a significant decline in pHVA over time. Interestingly, Green and coworkers (15) found similar changes in plasma dopamine during haloperidol treatment, although no early surge was detected as there was with pHVA. Although Meltzer (20) found no significant difference when comparing pHVA levels between unmedicated schizophrenic patients and normal controls, clozapine treatment in this study also was associated with a resulting decrease in pHVA. Bowers et al. (5) established a significant relationship between pretreatment pHVA levels and early neuroleptic treatment response in psychotic patients. Likewise Mazure et al. (19) determined that in a group of patients treated with perphenazine, pretreatment pHVA was significantly correlated with posttreatment outcome. In that study, as in the study by Chang et al. (6), good responders had higher mean levels of pHVA than did nonresponders. Additionally, Davidson et al. (8) also showed that patients who have neuroleptics discontinued and develop symptom exacerbation, experience an increase in pHVA. Davila and Friedhoff (10), determining both pHVA and prolactin, reported that the ratio of prolactin to pHVA had stronger correlations with clinical outcome than pHVA alone. Considering all of the studies, most of which are concordant in major detail, there is little question that one or more aspects of the pHVA response to neuroleptics is a predictor of the therapeutic response. The mechanism of this effect, however, remains a matter for speculation, debate, and disagreement.

It is now generally agreed that the response of pHVA to the initiation of neuroleptic treatment follows a typical pattern. Neuroleptics are potent D2 antagonists, blocking signal transduction to postsynaptic D2 receptors, and interfering with the sensing function of presynaptic D2 receptors (for review, see ref. 13). The presynaptic neurons, not sensing excess dopamine in the synaptic cleft, release massive amounts of this transmitter into the synaptic space. The high concentration of dopamine impinging on the blocked postsynaptic receptors competes with the neuroleptic for the binding site; however, the neuroleptic antagonist has a much higher affinity than dopamine for the postsynaptic D2 receptor and thus is probably only minimally displaced. Signal transduction, therefore, remains largely disrupted at early stages of treatment.

Continued interference with neurotransmission invokes another compensatory response—the appearance of new D2 receptors on the postsynaptic neuron (for review, see ref. 14). The density of these receptors can increase by as much as 25% or 30%; however the density in the postsynaptic region of the neuron could be substantially higher. If treatment is continued and blockade is maintained, ultimately the increased release of dopamine from presynaptic terminals returns to baseline levels or below through a process called depolarization blockade (22). The mechanism for this phenomenon is not well understood; however, it is accompanied by a reduction in the activity of the rate-limiting enzyme in dopamine synthesis, tyrosine hydroxylase (17), just as the initial increase in dopamine release was accompanied by an increase in tyrosine hydroxylase activity.

This sequence of events, which has been shown to occur at dopaminergic synapses in the brain, is paralleled almost exactly by changes in pHVA: (a) the initial increase in pHVA that follows initiation of neuroleptic treatment parallels the initial increase in dopamine release from presynaptic terminals, (b) the return of pHVA to baseline levels or below parallels the increase in the number of postsynaptic receptors and the reduction in dopamine release and synthesis as a result of depolarization blockade. Nonetheless serious questions remain as to whether these central events are responsible for the changes in the HVA found in plasma.

A major problem in understanding the mechanism of the pHVA response is the fact that most pHVA does not come from the central nervous system (CNS). It is almost certain, however, that neuroleptics work via their effects on central dopaminergic neurons. Dopamine has several roles in the body. It acts as a neurotransmitter in a network of dopaminergic neurons in the prefrontal cortex and in a number of interacting subcortical nuclei. It also acts on peripheral dopamine receptors principally in the vascular system and kidneys. Dopamine has another important role as a precursor of norepinephrine. To meet kinetic requirements for the synthesis of norepinephrine, dopamine must be produced in excess from its precursor L-dopa. An additional factor to be taken into consideration is that dopamine cannot cross the blood–brain barrier in either direction. Thus all dopamine in the brain is made from L-dopa, which can cross the barrier, and all dopamine found in the periphery originates in the periphery.

The HVA found in circulating plasma thus, comes from (a) dopamine released from central dopaminergic neurons and subjected to the action of monamine oxidase and catechol-O-methyltransferase, (b) dopamine acting as a precursor in the biosynthesis of norepinephrine in noradrenergic neurons in the brain, (c) dopamine released from peripheral dopaminergic neurons, (d) dopamine acting as a precursor in peripheral noradrenergic neurons, or (e) dopamine originating as a metabolite of L-dopa in any tissue in which 1-amino acid decarboxylase, the enzyme converting L-dopa to dopamine can be found. Finally the level of HVA in plasma is determined not only by its rate of synthesis but also by its rate of excretion, primarily in urine (16). The net result of all of this is that most pHVA does not originate in the brain or even from dopamine acting as a neurotransmitter.

These observations led us to ask the question, How can changes in a metabolite that comes mainly from sources other than from dopamine acting as a transmitter so successfully predict the outcome of a treatment that acts by modifying dopaminergic transmission? To attempt to understand this, we reviewed an experiment we carried out in 1978 (2). In this study, we used therapeutic response to haloperidol and resting finger tremor as outcome measures. Resting finger tremor reflects the continuous microoscillation of the fingers. This oscillation probably increases finger dexterity by making it possible to initiate a finger movement with a running start. The tremor is under strong dopaminergic control. Neuroleptic drugs, which are D2 antagonists, increase tremor amplitude and slow its frequency. Parkinsonism, a hypodopaminergic syndrome, also has this effect. Dopaminergic agonists, on the other hand, decrease amplitude and increase frequency. This tremor is largely under the control of spinal, not brain, neurons.

In our experiment, carried out years before pHVA could be assayed by high-performance liquid chromatography (HPLC), we measured baseline finger tremor in patients with schizophrenia, then began treatment with haloperidol. We proposed that subjects in whom neuroleptic administration had little effect on tremor released large amounts of dopamine which overcame the blockade. Four days after haloperidol treatment was begun we again measured tremor. Those subjects who best overcame the neuroleptic blockade, presumably by a large compensatory release of dopamine, had the least increase in amplitude and slowing of frequency. We then continued haloperidol treatment for 28 days and measured improvement in mental status. The subjects showing the least change in either tremor amplitude or frequency on the 4th day of the haloperidol treatment had the best therapeutic response to the haloperidol. Subjects with the most vigorous release of dopamine on the 4th day of haloperidol treatment would tend to overcome the haloperidol blockade of D2 receptors and not suffer a change in tremor. Those with a sluggish release, on the other hand, would have slowing of tremor frequency and increased amplitude.

Somewhat to our surprise those subjects with evidence of preexisting low dopaminergic activity did not do well therapeutically, whereas those with evidence of an active dopaminergic system did. This seeming paradox is addressed later in this chapter. It should be noted that haloperidol treatment tends to reduce dopaminergic activity in responders (as reflected by pHVA) to the point it was before treatment in nonresponders.

Despite the fact that plasma HVA is a peripheral dopaminergic measure, it is at least as good a predictor as HVA found in cerebrospinal fluid. Van Kammen (24) reviewed the literature and theorized that dopaminergic turnover in certain regions of the brain (such as cortical areas) are greater than in other brain regions. As a result, measurements of CSF HVA may predominantly represent dopaminergic turnover in cortical areas, rather than reflecting what transpires in specific striatal dopaminergic regions. Interestingly, Davidson and colleagues reported that the results of a study utilizing debrisoquin provided preliminary evidence correlating the relationship between pHVA and central HVA. Thus measuring pHVA may represent a practical approach to estimating dopamine turnover in the human brain.

One plausible explanation as to why pHVA changes, in response to neuroleptic treatment, is that neuroleptics have a correlated effect on the diverse sources of pHVA. If a number of different pools of origin of HVA contribute to the plasma pool in a correlated manner, changes in pHVA could be highly predictive of therapeutic response, even though it does not reflect only that fraction of pHVA that originates in presynaptic terminals of relevant brain neurons.

Surprisingly little has been reported about the relationship of pHVA changes during neuroleptic treatment and the emergence of side effects, in particular extrapyramidal side effects. Early in the use of neuroleptics it was proposed that the emergence of Parkinsonianlike tremor and rigidity was a good therapeutic indicator (1). Inasmuch as extrapyramidal side effects (EPS) result from low dopaminergic activity, and neuroleptics reduce dopaminergic activity, it was felt that EPS reflected the fact that hypodopaminergia was achieved. However, EPS has not turned out to be a good predictor of therapeutic outcome, perhaps because it occurs, initially, at relatively low levels of neuroleptic blockade, often before effective doses are achieved.

One complication in studying predictors of therapeutic response is the definition of a responder. As more data have been accumulated, a more specific definition of responders and nonresponders has begun to emerge. From clinical studies, it has been found that two types of manifestations of schizophrenia cluster together. These have been variously named but are now generally called positive and negative symptom types of schizophrenia. Positive symptoms are those generally found in more acute patients or in those having an acute recurrence and include delusions, labile affect, auditory hallucinations, and disturbances in associative processing. Negative symptoms include social withdrawal, flat affect, and anhedonia (for a complete list of negative symptoms see ref. 3). Few patients have a pure distillate of one type of symptom or the other; thus the characterization of a patient as positive or negative generally refers to the dominant symptom type. There are a number of studies showing that positive-symptom patients respond better to treatment with typical neuroleptics than negative symptom types, but pHVA has been measured in only a few of these studies (7, 11, 21). In a study by Davila et al. (11), it has been found that negative symptom patients do not have the initial increase in pHVA associated with the beginning of neuroleptic treatment, nor do they have the later decline in pHVA. Also there have been several reports that nonresponders have lower baseline pHVA, although these have not been rated for positive or negative symptoms (4, 6, 19, 23, 25).

Flat affect has been considered to be the sine qua non of the negative subtype. This is of interest because of the flat affectual responses of Parkinson's patients who have a central hypodopaminergic syndrome, and because neuroleptics, which reduce dopaminergic activity pharmacologically, produce blunted or flat affect in many cases. There is, therefore, reason to believe that affect is under strong dopaminergic regulation. This conclusion can also be supported by data obtained from observations of patients with extreme affective lability, as seen in hypomania, for example. In these patients, treatment with neuroleptic drugs frequently reduces affectivity. Thus we have proposed that a reduction of central dopaminergic activity may produce blunting of affective responses and that affect is under dopaminergic control. If this is the case, then one could conclude that negative-symptom patients have low central dopaminergic activity.

Affect may also play a central role in regulating thinking processes. There may be an association between flat affect and concrete thinking and between labile affect and looseness of associations. In mania, for instance, associations are made between thoughts that are related in nonsubstantive ways, because they sound alike, for example, or rhyme with each other. At one end of the affect spectrum, flatness is associated with concrete thinking in which a conceptual or symbolic dimension is missing. On the other end in a highly labile affective state, relationships are seen between thoughts, based on trivial connections, characterized as flight of ideas. Creative thinking, which involves making new associations, may lie somewhere between these two extremes and be dependent on more modest affectivity.

This chapter is not about thinking or affect, thus only scant attention can be given to those subjects. A consideration of these topics is necessary, however, to begin to address questions concerning the mechanism of the pHVA response. If overactivity of the dopaminergic system were the primary etiology of schizophrenia, as was proposed in the early dopamine hypothesis (18) then it would not be possible to understand patients with negative symptoms who have clinical evidence of low dopaminergic activity (flat affect) confirmed by generally low pHVA in these subjects. On the other hand if dopamine plays a role in the pHVA response only as a mediator of the treatment response, then it may be possible to reconcile pHVA changes with clinical response.

A new element has been introduced by the widespread use of clozapine, and the experimental use of many so-called atypical drugs. It is reported that these compounds can be used to effectively treat many patients with negative symptoms (low pHVA). These patients frequently have low baseline pHVA, which is not affected by treatment with typical neuroleptics, nor does their clinical state improve significantly. What is perplexing is the fact that responders (generally positive-symptom type) have a decrease in pHVA only to the baseline level of poor outcome patients (4). Do atypical drugs, more particularly clozapine, reduce pHVA further in negative-symptom patients than typical neuroleptics? This question is being pursued by several groups but cannot yet be answered definitively.

We have formulated a hypothesis in which we attempt to reconcile observations of the pHVA response with well-established clinical observations of patients with schizophrenia and their response to treatment. The proposal that best fits the existing facts is that the dopaminergic system in the brain is a restitutive or compensatory system. In the face of inescapable stress of either internal or external origin, of such nature as to threaten the integrity of the thinking function, the dopaminergic system turns itself down, resulting in affectual blunting, an increase in concrete thinking, and through a poorly understood mechanism, a decrease in pHVA. Patients with positive- or negative-type schizophrenia would carry a schizophrenia gene, but not necessarily the same one. In negative-symptom patients, a compensatory decrease in dopaminergic activity would have taken place resulting in affectual blunting and an increase in concrete thinking, but that did not overcome the deficit produced by the schizophrenia gene. Thus these patients would remain incapacitated but with the addition of flat affect and concrete thinking as manifestations of the invocation of a decrease in dopaminergic activity resulting from the action of the compensatory system. Flat affect and concrete thinking, in this proposal, are not manifestations of schizophrenia, but result from the turning down of the dopaminergic compensatory system. It should be noted that normal subjects, in response to intractable stress, also often manifest blunted affect.

We have proposed that positive patients have two defects: (a) In one a schizophrenia gene or a defect in a gene is involved in regulating activity of the dopaminergic compensatory system. The dopaminergic system in these patients cannot turn itself down but can be turned down pharmacologically by neuroleptics. (b) Negative-symptom patients, by contrast, get little benefit from neuroleptics, because their dopaminergic system has reduced itself spontaneously; thus typical neuroleptics have little additional effect. One important challenge of this hypothesis will be to see if clozapine can reduce pHVA more than typical neuroleptics do in those patients who respond to this drug.

Measurement of pHVA over only a few days of neuroleptic treatment makes it possible to predict the therapeutic outcome of the treatment. It is not clear why this works as a predictor because much of the HVA in plasma does not originate in the brain or even from dopamine released during neurotransmission. Further understanding of this mechanism may clarify the physiological role of the dopaminergic system in maintaining mental stability.

This work was supported in part by Grant MH 35976 and Research Scientist Award MH 14024, both from the National Institute of Mental Health.

published 2000