should schizophrenia be treated as a neurocognitive disorder?

Should Schizophrenia Be Treated as aNeurocognitive Disorder?

by Michael Foster Qreen and Keith H. Nuechterlein

Abstract

The search is on for meaningful psychopharmacologi-cal and cognitive/behavioral interventions for neu-rocognitive deficits in schizophrenia. Findings in thisarea are emerging rapidly, and in the absence of inte-grating frameworks, they are destined to emergechaotically. Clear guidelines for testing neurocognitiveinterventions and interpreting results are critical atthis early stage. In this article, we present three mod-els of increasing complexity that attempt to elucidatethe role of neurocognitive deficits in schizophrenia inrelation to treatment and outcome. Through discus-sion of the models, we will consider methodologicalissues and interpretive challenges facing this line ofinvestigation, including direct versus indirect neu-rocognitive effects of antipsychotic medications, selec-tion of particular neurocognitive constructs for inter-vention, the importance of construct validity ininterpreting cognitive/behavioral studies, and theexpected durability of treatment effects. With a grow-ing confidence that some neurocognitive deficits inschizophrenia can be modified, questions that seemedirrelevant only a few years ago are now fundamental.The field will need to reconsider what constitutes asuccessful intervention, what the relevant outcomesare, and how to define treatment efficacy.

Keywords: Neurocognition, cognition, antipsy-chotic medication, cognitive remediation, functionaloutcome.

Schizophrenia Bulletin, 25(2):309-318,1999.

Although the presentation of schizophrenia has changedminimally over time, our perceptions of the disorder havechanged dramatically, particularly over the past twodecades. During the 1980s, schizophrenia went frombeing viewed as mainly a progressively deteriorating dis-order to one of neurodevelopmental origin, partly becauseit was discovered that very early events can influence risk

for the disorder. Also during the 1980s, the predominantview of the phenomenology of schizophrenia broadenedbeyond a narrow focus on psychotic symptoms to includenegative symptoms as well. In the 1990s, another changein our perception of schizophrenia occurred, which maybe even more fundamental. With this change, we haveexpanded the phenomenology of schizophrenia even fur-ther, beyond symptoms altogether, to include a strongemphasis on neurocognitive aspects of schizophrenia.

The study of the neurocognition of schizophrenia isnot new; key deficits in attention, perception, and cogni-tion were noted by very early descriptive psychopatholo-gists (Bleuler 1950; Kraepelin 1913/1971), and experi-mental psychopathologists have worked to identify thecore abnormalities for several decades (Shakow 1962;Chapman and Chapman 1973; Goldstein 1978;Nuechterlein and Dawson 1984; Braff 1993). But neu-rocognitive studies of schizophrenia are now viewed ashaving direct implications for treatment. Interest in thetopic has extended far beyond a rather small and dedi-cated group of experimentalists. Practitioners nowbelieve that schizophrenia can legitimately be viewed, inessence, as a disorder of neurocognition.

Articles in this theme issue represent the state of theart in pharmacological and cognitive/behavioral treat-ments for neurocognitive deficits in schizophrenia.Findings in this area are emerging rapidly and in theabsence of integrating frameworks are destined to emergechaotically. In this article, we present three models ofincreasing complexity that are designed to serve an orga-nizing function. Through discussion of die models, wewill address methodological challenges inherent inresearch on neurocognitive interventions, drawing fromthe articles in this dieme issue to illustrate points. Let'sstart with the simplest model.

Reprint requests should be sent to Dr. M.F. Green, UCLA-Neuropsychiatric Institute, 760 Westwood Plaza, C9-420, Los Angeles,CA 90024-1759.

309

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article/25/2/309/1919077 by guest on 01 January 2022

Schizophrenia Bulletin, Vol. 25, No. 2, 1999 M.F. Green and K.H. Nuechterlein

Simple Model

Figure 1 depicts a simple model based entirely on threeconclusions that can be drawn from the literature. First,conventional neuroleptics are generally effective for psy-chotic symptoms, but their effects on neurocognition arerelatively weak (Cassens et al. 1990; Strauss 1993).Conventional neuroleptics sometimes have short-termdetrimental effects on tests of psychomotor speed, andsome studies have shown longer term benefits for vigi-lance and early visual processing. However, the modalfinding across studies and neurocognitive constructs isthat of no significant effect.

Second, the cross-sectional relationships between neu-rocognition and psychotic symptoms are usually minimal,particularly for hallucinations and delusions. The relation-ships are weak for most neurocognitive constructs, evenwhen they are statistically significant (Comblatt et al.1985; Green and Walker 1985; Nuechterlein et al. 1986;Green et al. 1992; Strauss 1993). In cross-sectional stud-ies, it is unusual to find relationships between psychoticsymptoms and neurocognitive measures in which morethan 10 percent of the variance can be explained (Goldberget al. 1993). Nevertheless, it is possible that some veryspecific neurocognitive processes underlie particular posi-tive symptoms. For example, several rather comprehen-sive theoretical models have proposed testable linksbetween particular types of psychotic symptoms and neu-rocognitive deficits (Frith and Done 1988; Hemsley 1994).In addition, some studies have linked formal thought disor-der (sometimes considered to be on a disorganized symp-tom dimension instead of a positive symptom dimension)to performance on a highly specific laboratory measure of

Figure 1. Pathways from conventional antipsy-chotic treatment to functional outcome

Neurocognition

Conventional IAntipsychotlcsl

FunctionalOutcome

PsychoticSymptoms

Note.—This simple model represents a possible mismatch: theaspect of illness most influenced by conventional antipsychoticmedications (psychotic symptoms) is not the aspect most associ-ated with functional outcome.

semantic priming (Spitzer 1997). While the possibility ofinformative specific linkages remains, our position is thatmodels will have greater explanatory power when psy-chotic symptoms and neurocognition are treated as sepa-rate domains. While psychotic symptoms and neurocogni-tive processes are not completely separate, they aresufficiently separate. Further, we know that some neu-rocognitive deficits are entirely separate in terms of timecourse, because they remain unchanged when psychoticsymptoms improve (Comblatt et al. 1997) or even fullyremit (Nuechterlein et al. 1992).

Third, initial studies suggest that certain neurocogni-tive deficits are rather good predictors or correlates offunctional outcome, whereas the relationships betweenpsychotic symptoms and functional outcome are weak orquestionable. In a review of the literature (Green 1996),mostly from 1990 to 1995, certain neurocognitive con-structs such as verbal memory and vigilance emerged asreliable correlates and predictors of several outcome areasin chronic schizophrenia, including community function-ing, social problem solving, and psychosocial skill acqui-sition. Within these same studies, the relationshipsbetween psychotic symptoms and functional outcomewere much weaker than those for neurocognitive vari-ables. Data from Bellack et al. (1999, this issue), as wellas several recent studies published subsequent to thisreview, lend support for these conclusions (Dickerson etal. 1996; Brekke et al. 1997; Harvey et al. 1997; Velliganetal. 1997).

A problematic mismatch is immediately apparent inthe simple model. Conventional neuroleptics have animpact on symptoms but, in general, not on most areas ofneurocognition. However, functional outcome appears tobe more closely related to neurocognitive abilities than tosymptoms. When viewed within this model, commontreatment experiences that previously seemed paradoxicalno longer are. A treatment team should not be puzzledwhen it successfully eliminates psychotic symptoms of apatient only to find that the patient is unable to resumesocial functioning. Likewise, parents should not be sur-prised to discover that their son or daughter with schizo-phrenia is able to return to work despite rather prominentauditory hallucinations. These situations would beexpected when viewed within the simple model in figure 1.

Which Neurocognitive Deficits to Target If neurocog-nitive deficits are more closely related to functional out-come, then should neurocognitive deficits themselvesbecome a target of treatment? Improvement in neurocog-nition suggests but does not logically guarantee improvedoutcome, because we do not yet understand the causalpathways. Most of the articles in this theme issue consider

310

Dow



Neurocognitive Deficits Schizophrenia Bulletin, Vol. 25, No. 2, 1999

the prerequisite question of whether it is possible to alterneurocognition in schizophrenia, and the results are gener-ally encouraging. Assuming we can improve neurocogni-tive deficits, we are faced with a difficult decision: Withsuch a wide array of deficits to choose from, which onesdo we select for intervention efforts? The need for selec-tion is particularly pronounced for cognitive/behavioralinterventions that focus on one neurocognitive deficit (andmaybe one cognitive construct) at a time. However, phar-macological intervention studies are not immune to thisproblem, because it is highly likely that certain agents willinfluence some neurocognitive constructs more than others(Meltzer and McGurk 1999, this issue). One way toapproach this question of construct selection is first todetermine which deficits are associated with outcomeareas of interest. For example, if the goal is to improveskill acquisition, constructs that have reliable associationswith this outcome area, such as verbal memory and vigi-lance, may be excellent candidates for interventions.

An alternative approach would be to target deficitsfor which the neural substrates are known. This approachmight be particularly attractive for pharmacological inter-vention studies, especially to the degree that certain neu-rocognitive deficits may be linked to specific neurotrans-

mitter systems in particular brain regions (Robbins andEveritt 1995; Keefe et al. 1999). Working memory, forexample, has substantial current appeal as a target forintervention, partially because of our understanding of itsneural circuitry and neurotransmitter mediation (Fuster1989; Goldman-Rakic 1991).

More Complex Model

While the simple model in figure 1 has the advantage ofparsimony, it is woefully incomplete. A more complexmodel, depicted in figure 2, has three additional compo-nents: new antipsychotic agents, anticholinergic agents,and negative symptoms. Three types of arrows have beenadded to convey the presumed strengths of the associa-tions. While not based on a formal path analysis, themodel is intentionally presented in the style of a path dia-gram. The neurocognition component is placed in an ovalfor emphasis, not to indicate that it is a separate type ofvariable.

Whereas the impact of conventional antipsychoticmedications on neurocognition has been unimpressive,early indications are that the story for newer antipsychotic

Figure 2. Neurocognition and functional outcome: Individual factors

NovelAntipsychoticMedications

AnticholinergicMedications

ConventionalAntipsychoticMedications

NeurocognitionFunctionalOutcome

PsychoticSymptoms

NegativeSymptoms

AssociationsStrongModerate

->• Potential(+) Beneficial Effects(•) Deleterious Effects

Note.—This more complex model includes boxes for novel as well as conventional antipsychotic medications, negative as well as positivesymptoms, and adjunctive anticholinergic medications. The model emphasizes individual factors as opposed to psychosocial, familial,and community determinants.

311

Dow




Figure 3. Neurocognition and functional outcome: Possible mediators

Cognitive/BehavioralInterventions I

AdjunctivePharmacology

Individual Factors

BasicNeurocognition

Novel - 7 ^^Antipsychotic | — r ^ Aeconda^yXMedications I ( + M( verbal

emory^^Executlve

AntichohnergicMedications

ConventionalAntipsychoticMedications

PsychoticSymptoms

SocialCognition

NegativeSymptoms

EmotionPerceptionSocialSchemaCopingSkillsInsight

FunctionalOutcome

• Social

• Occupational

• PatientSatisfaction

• CaregiverBurden

( • )

AssociationsStrong

> Moderate>- Potential

Beneficial EffectsDeleterious Effects

Note.—This complex model shows differentiated sub-components of basic neurocognition and functional outcome. It includes cognitive/behavioral and adjunctive pharmacology as additional interventions, and social cognition as a possible mediator between basic neu-rocognition and functional outcome.

medications is more encouraging (Hagger et al. 1993;Green et al. 1997; Jeste et al. 1998). Because this field ofinvestigation is still young, the arrow in the model fromnew antipsychotic medications to neurocognition indi-cates a potential, instead of a known, effect. Three arti-cles in this issue consider the role of new antipsychoticagents for treatment of neurocognitive deficits (Keefe etal. 1999; Kern et al. 1999, this issue; Meltzer and McGurk1999, this issue). These articles contribute to an emergingopinion that the new medications are better than the oldones for neurocognition. If this is so, we need to considerwhether the beneficial effects of these medications aredirect or indirect.

Direct versus Indirect Psychopharmacological Effects.A direct effect would involve an action of a particularmedication on a particular neurocognitive construct. An

indirect effect would mean that some aspect of treatmentother than a direct action of the agent itself is responsiblefor a change in neurocognition. A possible mechanism foran indirect effect is represented in figure 2. Consider thesituation in which a new antipsychotic agent is comparedwith a conventional one. Conventional antipsychoticmedications involve a much greater coadministration ofanticholinergic medications (e.g., benztropine mesylate)than do novel medications. Medications with strong anti-cholinergic properties are known to have a negative effecton certain aspects of neurocognition, particularly memory(Spohn and Strauss 1989). Thus, it is unclear whether adifferential treatment effect is the result of somethinggood (from the new medication) or the absence of some-thing bad (from the anticholinergic agent).

Although anticholinergic agents have a reputation fordisrupting neurocognition (Spohn and Strauss 1989), we

312

Dow




know surprisingly little about the degree and scope of thedetrimental effect. Data indicate a negative effect onaspects of secondary verbal memory that may rely onrehearsal strategies. In contrast, other aspects of memory,such as immediate or working memory, are less affected(Drachman and Leavitt 1974; Sweeney et al. 1991), andthe effects on other neurocognitive abilities, such as per-ception, are relatively unknown. Hence, we do not knowif anticholinergic agents are the neurocognitive culpritswe often believe them to be. Nonetheless, these medica-tions still represent the source of a possible indirect effectOne way to control for this possibility is demonstrated inthe article by Kem et al. (1999, this issue), in which anti-cholinergic medication was entered as a covariate into thestatistical model. This made it possible to demonstratethat the neurocognitive effects of the antipsychotic agentwere significant, over and above any effect of the anti-cholinergic medication. The tentative conclusion is thatnew antipsychotic medication may actually be good forneurocognition, instead of merely not bad.

How does one address and interpret the role of symp-tom changes and neurocognition? One variable in thearticle by Kern et al. (1999, this issue), recall consistency,showed a significant differential response to risperidonecompared with haloperidol. This beneficial effectremained significant when anticholinergic medication wasentered into the model. However, the effect weakenedslightly and became a trend when changes in psychoticand negative symptoms were entered. Do we concludethat the neurocognitive effect of the medication is an indi-rect effect of changes in clinical state? Not necessarily. Itdepends on how we view the interrelationships betweensymptoms and neurocognition. If neurocognitive changescontributed to the symptom changes, statistical adjust-ment for the presumed "effect" of symptom changeswould lead to the entirely incorrect interpretation that themedication's impact on neurocognition is accounted forby its impact on symptoms. Meehl (1971) describes simi-lar interpretive errors that can arise when effects of a vari-able X are used as covariates to examine the relationshipsbetween X and Y. To disentangle such relationships, care-ful examination of temporal sequence and of the interven-tions impact on neurocognition in relatively asympto-matic patients will likely be needed.

Alternative Causal Pathways for Negative Symptoms.The addition of a component for negative symptoms, sep-arate from psychotic symptoms, creates new pathwaysand new challenges. One question is the degree of over-lap between neurocognitive deficits and negative symp-toms. Relationships tend to be stronger between neu-rocognitive deficits and negative symptoms (Nuechterlein

et al. 1986; Censits et al. 1997) or between neurocognitivedeficits and the more narrowly defined deficit syndrome(Buchanan et al. 1997) than they are between neurocogni-tive deficits and psychotic symptoms. However, since thepercent of variance explained is still relatively small(10%—15%), we believe that negative symptoms and neu-rocognitive deficits can be placed on different pathways.Also, neurocognitive deficits appear to start earlier thannegative symptoms (Cornblatt et al. 1992). We have con-servatively used a double-headed arrow between neu-rocognition and negative symptoms in figure 2 to indicateshared variance without making any assumptions aboutcausality, but it is entirely possible that the critical path-way runs from neurocognitive deficits to negative symp-toms. In support of that view, Nuechterlein et al. (1986)used longitudinal data to conclude that vigilance and per-ceptual span deficits act as vulnerability factors for devel-opment of negative symptoms rather than as secondaryeffects of negative symptoms.

A single-headed arrow indicates a moderate relation-ship between negative symptoms and functional outcome,but the nature of this relationship is not yet clear. First,the assessment of negative symptoms sometimes overlapswith the assessment of social functioning, so it is notalways clear if relationships are due to associationsbetween separate constructs or to redundancy of assess-ment. Assuming the constructs are moderately associated,the relationships between negative symptoms and func-tional outcome can have two quite different interpreta-tions. It might be that negative symptoms have a direct,causal impact on functional outcome, an interpretationthat is obvious, intuitive, and quite possibly wrong.Instead, the moderate relationship may be explained by(1) the shared variance between negative symptoms andneurocognition and (2) the strong associations betweenneurocognition and functional outcome. In other words,the causal pathways from negative symptoms to func-tional outcome may be due to a common neurocognitiveintersection. Indeed, two recent studies based on statisti-cal models (Harvey et al. 1997; Velligan et al. 1997) havearrived at exactly this conclusion.

Complex Model

The second model is also incomplete, so we introducethree new components in a third and final model: cogni-tive/behavioral interventions, adjunctive pharmacology,and social cognition (see figure 3). To distinguish theneurocognitive oval from the box on social cognition, theoval was relabeled as "basic" neurocognition. This modelalso depicts some of the subcomponents of the complex,

313

Dow




multifactorial domains of neurocognition and functionaloutcome.

Three articles in this theme issue address cognitive/behavioral interventions for neurocognitive deficits(Bellack et al. 1999, this issue; Spaulding et al. 1999, thisissue; Wykes et al. 1999, this issue). One key interpretivechallenge for many cognitive/behavioral studies is con-struct validity.

Construct Validity. To understand construct validity,one needs to distinguish between a construct (or latentvariable) and an indicator (Loehlin 1987; Nunnally 1978).In the study of neurocognition, we are inevitably inter-ested in constructs that cannot be directly observed.Instead we measure performance on an indicator that pre-sumably reflects a particular construct. For example, per-formance on a continuous performance test is an indica-tor; vigilance is the underlying construct. Theneurocognitive function, vigilance, has numerous perfor-mance and psychophysiological indicators (Davies andParasuraman 1982; Nuechterlein 1991).

This distinction between the construct and the indica-tor creates difficulties in the context of remediation stud-ies. Consider a training method for performance on thecontinuous performance test: if subjects' performance onone task improves following training, it is difficult toknow if the intervention affected the underlying construct(vigilance) or some aspect of the task unrelated to vigi-lance (perhaps skill in initiating a button press). One wayto test for construct validity is to use multiple indicatorsof the same construct. When it comes to training schizo-phrenia patients on a continuous performance task, theresults have been mixed (Benedict et al. 1994; Medalia etal. 1998). Training on one task can bring about clearimprovement, but the benefits do not always extend toanother measure of the same construct. This situationcould arise if training improved the underlying constructbut the second test was not a good indicator of that con-struct. However, it is also possible that the trainingbrought about improvement at the level of the indicator,but not at the level of the construct.

Construct validity is somewhat less of a concern forpharmacological studies than it is for cognitive/behavioralinterventions that focus training on particular tasks.Certainly, there is still the inherent separation of the con-struct from the measure, but with pharmacological inter-ventions, there is no training task. If a drug improves per-formance on a test of verbal working memory in acontrolled trial, it is often reasonable to conclude that thedrug helped verbal working memory. But without differ-ent indicators of verbal working memory, it remains pos-sible that the improvement is due to a pharmacological

effect on an isolated component of the verbal workingmemory task (e.g., oral fluency) rather than on functionsthat are considered to be more central to the construct.

The complex model includes two other componentsthat represent largely unexplored territory. One is adjunc-tive pharmacology; the other includes hypothesized medi-ators between neurocognition and functional outcome.

Adjunctive Psychopharmacology. The use of adjunc-tive pharmacology for neurocognitive deficits has onlyrecently received serious consideration (Davidson andKeefe 1995). Some of this emerging interest in adjunc-tive medications undoubtedly stems from recent develop-ments in the pharmacological treatment of cognitivedecline in dementia. The medications for dementia act onthe cholinergic system, which is related to memory func-tioning. It is not yet known whether this system is alsocritical to the pathophysiology of schizophrenia. Perhapsmore relevant to schizophrenia is the glutamate system. Ithas been suggested diat hypofunction of the A^-methyl-r>aspartate subtype of glutamate receptor may be a key dis-ease mechanism for schizophrenia (Olney and Farber1995). This receptor is modulated by glycine, which hasled to interest in using glycinergic agents to treat schizo-phrenia. Preliminary data on a small sample of patientsshowed improvement in choice reaction time when d-clycloserine, a partial glycine agonist, was added to con-ventional antipsychotic medications (Goff et al. 1995). Ifadjunctive agents are found to improve neurocognition,schizophrenia patients may start receiving one medicationfor each major domain of illness: symptoms and neu-rocognitive deficits.

Social Cognition as Mediator. Between basic neu-rocognition and functional outcome are a cluster of medi-ating variables labeled as social cognition. In the model,components of social cognition include emotion percep-tion, social schema, insight into illness, and coping/attri-butional strategies. Our intention is to distinguish thesevariables, which have substantial social or emotionalcomponents, from basic neurocognition, although the pre-cise boundaries between basic and social cognition are notclear-cut.

One key aspect of social cognition is the ability toperceive emotion in others. Perception of emotion inschizophrenia is associated with basic neurocognition(Schneider et al. 1995; Bryson et al. 1997), suggestingthat intact basic neurocognition is necessary for accurateperception of emotion. It is reasonable to expect that areduced ability to perceive emotion in others would resultin misinterpretation and compromised social interactions.Indeed, perception of emotion appears to be closely

314

Dow




related to social competence in schizophrenia inpatients(Mueser et al. 1996; Perm et al. 1996).

In this model, we have also included insight and cop-ing in the "social cognition" box even though they are notgenerally considered to be prototypic features of socialcognition. A subsequent model may place them in a sepa-rate box for "attributional" or "metacognitive" constructs.The relationship between insight and neurocognition isnot at all clear. Most studies have reported that insight isrelated to better neurocognitive performance, particularlyon measures of executive functioning (Silverstein andZenvic 1985; Young et al. 1993; Lysaker and Bell 1994).However, other studies have not found such relationships(Cuesta and Peralta 1994; Dickerson et al. 1997).Disparity in these findings may stem from differences inthe definition and measurement of insight.

By placing social cognition between basic neurocog-nition and functional outcome, this model may help toilluminate an issue that has preoccupied investigators:whether deficits in social cognition are general rather thanspecific (reviewed in Penn et al. 1997). For example, sev-eral studies have evaluated whether emotion perceptiondeficits in schizophrenia are specific to emotional stimulior whether they are part of general perception deficits.While this question remains scientifically interesting, thedistinction between a specific emotion perception deficitand a general nonemotion perception deficit is not entirelyrelevant to the complex model. In this model, basic andsocial cognition are closely related but not identical com-ponents. Basic neurocognition is a prerequisite for socialcognition, and social cognition, in turn, is a prerequisitefor social functioning.

Overarching Questions

The models described here help to isolate componentsthat serve as intersections between treatment and out-come. Stepping back from the components, several broadquestions emerge. The questions are rather basic, whichis understandable given their recency. Pondering interpre-tation and implications of neurocognitive interventionswill be pointless until there is some agreement that inter-vention in this domain works. Now that there is growingconfidence that at least some neurocognitive deficits inschizophrenia can be modified, we are forced to confrontseemingly simple definitional matters. Only a few yearsago these questions seemed irrelevant; now they seemfundamental.

What Constitutes a Successful Intervention? Someneurocognitive deficits in schizophrenia are rather stableover time (Nuechterlein et al. 1992, 1994; Cornblatt et al.

1997), and it would be unreasonable to expect long-stand-ing deficits to improve permanently with a short-livedtreatment. So, how long should an effect last for an inter-vention to be considered successful? The answer seemsto depend on whether the intervention is psychopharma-cological or cognitive/behavioral. If a medicationimproves neurocognition while it is administered but notafter it is stopped, it is usually considered a success. If acognitive/behavioral intervention improves neurocogni-tive performance while the training is administered butnot after the intervention is stopped, it is usually consid-ered a failure. As Wykes et al. point out (1999, thisissue), this double standard complicates notions of whatconstitutes successful treatment Whether an interventionis a success depends on both implicit and explicit goals.The goal of psychopharmacological interventions is totreat; the goal of cognitive/behavioral interventions is toretrain. The complementarity of these two related butseparate goals provides a rationale for combining theapproaches. To the extent that psychopharmacologicaltreatments are successful, they are likely to open the doorfor effective retraining.

It is possible that the dichotomy is not absolutebetween the treating and retraining approaches to neu-rocognitive deficits in schizophrenia. Behavioral studiesmake it clear that learned skills can deteriorate over timeand that booster sessions are often called for. For psy-chopharmacological treatments, it is entirely possible thata short-term treatment could reregulate a dysregulatedneurochemical system and not require ongoing adminis-tration, or at least not at the same dosage. Hence, theeffects of retraining are not necessarily durable, and theeffects of medications are not necessarily transient.

What Are the Relevant Outcomes? When a reason-able intervention for a neurocognitive deficit is achieved,the question shifts to the selection of a relevant outcome.Specifically, what benefits do we expect to see? Fre-quently, the goal of clinically based cognitive interventionprograms is symptomatic improvement. Although thisseems like a reasonable goal, it presents an interpretivechallenge. The model in figure 3 shows no direct mecha-nism-for an effect on psychotic symptoms.- Instead, neu-rocognitive interventions might be more beneficial fornonpsychotic symptoms such as blunted affect and hostil-ity. This, in fact, may be the case (Brenner et al. 1990).

However, it is likely that neurocognitive interventionswill have their greatest impact outside the domain ofsymptoms. The rather strong relationships between neu-rocognitive deficits and functional outcome indicates thatwe might expect to see an effect here. One notion is thatthese deficits act as "neurocognitive rate limiting factors"

315

Dow




and restrict the functional adaptation of the patient (Green1996). Following intervention for neurocognitive deficits,this limitation is eased, allowing for more complete skillacquisition and functioning. If new learning is required toachieve the functional gains, we should expect a time lagbetween any improvement in neurocognitive functioningand measurable improvement in outcome.

The Delta Question. A key question, which we refer toas the "delta" question, is whether changes in neurocogni-tion translate into changes in functional outcome. Giventhe associations between neurocognition and functionaloutcome, it is reasonable to expect that changes in the twodomains would be associated. However, most studieshave examined whether level of neurocognitive perfor-mance is associated with functional outcome, not whetherchanges in neurocognitive functioning are directly linkedto changes in functional outcome.

Some data bearing on this point suggest that changesin the two domains may be linked. For example,Buchanan et al. (1994) found that changes in memoryover 1 year were correlated with changes in quality of lifein schizophrenia patients. Wykes et al. (1999, this issue)report that change in cognitive flexibility was related toimprovement in social functioning in a relatively shorttreatment trial (8 weeks). Spaulding et al. (1999, thisissue) report that improvement in card sorting was associ-ated with improved social competence, and that improve-ment in verbal memory was associated with increasedpsychosocial skill acquisition over the course of a 6-month intervention study. These studies provide very pre-liminary, but encouraging, support for the hypothesis thatchanges in the neurocognitive domain will result inchanges in functional outcome.

It may be even more informative to phrase the deltaquestion in quantitative terms: How much of a change inneurocognition is needed to bring about a meaningfulchange in an outcome area? In this way, the question canbe answered with cost/benefit considerations in mind.Knowing the amount of neurocognitive improvementrequired to make a clinically meaningful improvement indaily functioning for a given domain would enable us tobetter evaluate the practical utility of neurocognitive gainsthat may be possible through psychopharmacological orcognitive/behavioral intervention. Clearly, a key chal-lenge facing this entire area of investigation will be tounderstand the process through which short-term perfor-mance gains can be converted into long-term functionalbenefits.

How to Define Efficacy. Several articles in this themeissue strongly suggest that the new generation of antipsy-chotic agents acts on more than one domain of illness. If

so, then we are confronted with the problem of how todefine efficacy. Traditionally, efficacy has been defined interms of reduction of psychotic symptoms, but morerecently, the reduction of negative symptoms has beenadded to the definition. However, if the new medicationsact on multiple domains of illness, then symptom reduc-tion is simply too narrow a definition of efficacy (Green etal. 1997). Even our language betrays us; the very term"antipsychotic" may prove to be too narrow for the newgeneration of medications.

To the extent that new medications and innovativecognitive/behavioral interventions act in the neurocogni-tive domain, it may be more accurate to describe an inter-mediate goal of treatment as impairment reduction. Andif psychopharmacological and cognitive/behavioral neu-rocognitive interventions translate into functional gainsfor patients, then even impairment reduction may be toonarrow. In this case it may be more accurate to describethe eventual goal of treatment as disability reduction. Thearticles in this theme issue can be viewed as concertedefforts toward these goals of impairment and disabilityreduction.

References

Bellack, A.S.; Gold, J.M.; and Buchanan, R.W. Cognitiverehabilitation for schizophrenia: Problems, prospects, andstrategies. Schizophrenia Bulletin, 25(2):257-274, 1999.

Benedict, R.H.B.; Harris, A.E.; Markow, T; McCormick,J.A.; Nuechterlein, K.H.; and Asarnow, R.F. Effects ofattention training on information processing in schizo-phrenia. Schizophrenia Bulletin, 20:537-546, 1994.

Bleuler, E. Dementia Praecox or the Group of Schizo-phrenias. New York, NY: International Universities Press,1950.

Braff, D. Information processing and attention dysfunc-tions in schizophrenia. Schizophrenia Bulletin, 19:233-259, 1993.

Brekke, J.S.; Raine, A.; Ansel, M.; and Lencz, T. Neuro-psychological and psychophysiological correlates of psy-chosocial functioning in schizophrenia. SchizophreniaBulletin, 23:19-28, 1997.

Brenner, H.D.; Kraemer, S.; Hermanutz, M.; and Hodel,B. Cognitive treatment in schizophrenia. In: Straube,E.R., and Halweg, K., eds. Schizophrenia: Concepts,Vulnerability, and Interventions. New York, NY: Springer-Verlag, 1990. pp. 161-192.

Bryson, G.; Bell, M.; and Lysaker, P. Affect recognition inschizophrenia: A function of global impairment or a spe-cific cognitive deficit. Psychiatry Research, 71:105-113,1997.

316

Dow




Buchanan, R.W.; Holstein, C ; and Breier, A. The compar-ative efficacy and long-term effect of clozapine treatmenton neuropsychological test performance. BiologicalPsychiatry, 36:717-725, 1994.

Buchanan, R.W.; Strauss, M.E.; Breier, A.; Kirkpatrick,B.; and Carpenter, W.T. Attentional impairments in deficitand nondeficit forms of schizophrenia. American Journalof Psychiatry, 154:363-370, 1997.

Cassens, G.; Inglis, A.K.; Appelbaum, P.S.; and Gutheil,T.G. Neuroleptics: Effects on neuropsychological functionin chronic schizophrenic patients. Schizophrenia Bulletin,16:477^99, 1990.

Censits, D.M.; Ragland, J.D.; Gur, R.C.; and Gur, R.E.Neuropsychological evidence supporting a neurodevelop-mental model of schizophrenia: A longitudinal study.Schizophrenia Research, 24:289-298, 1997.

Chapman, L.J., and Chapman, J.P. Problems in the mea-surement of cognitive deficit. Psychology Bulletin,79:380-385, 1973.

Cornblatt, B.A.; Lenzenweger, M.F.; Dworkin, R.H.; andErlenmeyer-Kimling, L. Positive and negative schizo-phrenic symptoms, attention, and information processing.Schizophrenia Bulletin, 11:397-408, 1985.

Cornblatt, B.; Lenzenweger, M.F.; Dworkin, R.; andErlenmeyer-Kimling, L. Childhood attentional dysfunc-tion predicts social deficits in unaffected adults at risk forschizophrenia. British Journal of Psychiatry, 161(Suppl.18):59-64, 1992.

Cornblatt, B.; Obuchowski, M.; Schnur, D.B.; andO'Brien, J.D. Attention and clinical symptoms in schizo-phrenia. Psychiatric Quarterly, 68:343-359, 1997.

Cuesta, M.J., and Peralta, V. Lack of insight in schizo-phrenia. Schizophrenia Bulletin, 20:359-366, 1994.

Davidson, M., and Keefe, R.S.E. Cognitive impairment asa target for pharmacological treatment in schizophrenia.Schizophrenia Research, 17:123-129, 1995.

Davies, D.R., and Parasuraman, R. The Psychology ofVigilance. London, England: Academic Press, 1982.

Dickerson, R; Boronow, J.J.; Ringel, N.; and Parente, F.Neurocognitive deficits and social functioning in outpa-tients with schizophrenia. Schizophrenia Research,21:75-83, 1996.

Dickerson, F.B.; Boronow, J.J.; Ringel, N.; and Parente, F.Lack of insight among outpatients with schizophrenia.Psychiatric Services, 48:195-199, 1997.

Drachman, D.A., and Leavitt, J. Human memory and thecholinergic system. Archives of Neurology, 30:113-121,1974.

Frith, CD., and Done, D.J. Towards a neuropsychology ofschizophrenia. British Journal of Psychiatry, 153:437-443, 1988.

Fuster, J.M. The Prefrontal Cortex: Anatomy, Physiology,and Neuropsychology of the Frontal Lobe. New York,NY: Raven Press, 1989.

Goff, D.C.; Tsai, G.; Manoach, D.S.; and Coyle, J.T.Dose-finding trial for D-cycloserine added to neurolepticsfor negative symptoms in schizophrenia. AmericanJournal of Psychiatry, 152:1213-1215, 1995.

Goldberg, T.E.; Gold, J.M.; Greenberg, R.; and Griffin, S.Contrasts between patients with affective disorders andpatients with schizophrenia on a neuropsychological testbattery. American Journal of Psychiatry, 150:1355-1362,1993.

Goldman-Rakic, P.S. Prefrontal cortical dysfunction inschizophrenia: The relevance of working memory. In:Carroll, B.J., and Barrett, J.E., eds. Psychopathology andthe Brain. New York, NY: Raven Press, 1991. pp. 1-23.

Goldstein, G. Cognitive and perceptual differencesbetween schizophrenics and organics. SchizophreniaBulletin, 4:160-185, 1978.

Green, M.F. What are the functional consequences of neu-rocognitive deficits in schizophrenia? American Journalof Psychiatry, 153:321-330, 1996.

Green, M.F.; Marshall, B.D.; Wirshing, W.C.; Ames, D.;Marder, S.R.; McGurk, S.; Kem, R.S.; and Mintz, J. Doesrisperidone improve verbal working memory in treatment-resistant schizophrenia? American Journal of Psychiatry,154:799-804, 1997.

Green, M.F.; Nuechterlein, K.H.; and Gaier, D.J. Sus-tained and selective attention in schizophrenia. Progressin Experimental Personality and PsychopathologyResearch, 15:290-313, 1992.

Green, M.F., and Walker, E. Neuropsychological perfor-mance and positive and negative symptoms in schizophre-nia. Journal of Abnormal Psychology, 94:460-469, 1985.

Hagger, C ; Buckley, P.; Kenny, J.T.; Friedman, L.;Ubogy, D.; and Meltzer, H.Y. Improvement in cognitivefunctions and psychiatric symptoms in treatment-refrac-tory schizophrenic patients receiving clozapine. Bio-logical Psychiatry, 34:702-712, 1993.

Harvey, P.D.; Davidson, M.; Mueser, K.T.; Parrella, M.;White, L.; and Powchik, P. Social-adaptive functioningevaluation (SAFE): A rating scale for geriatric psychiatricpatients. Schizophrenia Bulletin, 23:131-139, 1997.

Hemsley, D.R. Perceptual and cognitive abnormalities asthe bases for schizophrenic symptoms. In: Anthony, D.S.,

317

Dow




and Cutting, J.C., eds. The Neuropsychology of Schizo-phrenia. Hove, UK: Lawrence Erlbaum Associates, 1994.pp. 97-116.

Jeste, D.V.; Lohr, J.B.; Eastman, J.H.; Rockwell, E.; andCaligiuri, M.P. Adverse neurobiological effects of long-term use of neuroleptics: Human and animal studies.Journal of Psychiatric Research, 32:201-214, 1998.

Keefe, R.S.E.; Silva, S.G.; Perkins, D.O.; and Lieberman,J.A. The effects of atypical antipsychotic drugs on neu-rocognitive impairment in schizophrenia: A review andmeta-analysis. Schizophrenia Bulletin, 25(2):201-222,1999.

Kern, R.S.; Green, M.F.; Marshall, B.D.; Wirshing, W.C.;Wirshing, D.; McGurk, S.R.; Marder, S.R.; and Mintz, J.Risperidone vs. haloperidol on secondary memory: Cannewer antipsychotic medications aid learning? Schizo-phrenia Bulletin, 25(2):223-233, 1999.

Kraepelin, E. Dementia Praecox and Paraphrenia[1913/1919]. Huntington, NY: Robert E. KriegerPublishing Co. Inc., 1971.

Loehlin, J.C. Latent Variable Models: An Introduction toFactor, Path, and Structural Analysis. Hillsdale, NJ:Lawrence Erlbaum Associates, 1987.

Lysaker, P., and Bell, M. Insight and cognitive impairmentin schizophrenia: Performance on repeated administra-tions of the Wisconsin Card Sorting Test. Journal ofNervous and Mental Disease, 182:656-660, 1994.

Medalia, A.; Aluma, M.; Tryon, W.; and Merriam, A.Effectiveness of attention training in schizophrenia.Schizophrenia Bulletin, 24:147-152, 1998.

Meehl, P.E. High school yearbooks: A reply to Schwartz.Journal of Abnormal Psychology, 77:143-148, 1971.

Meltzer, H.Y., and McGurk, S.R. The effects of clozapine,risperidone, and olanzapine on cognitive function inschizophrenia. Schizophrenia Bulletin, 25(2):233-255,1999.

Mueser, K.T.; Doonan, B.; Penn, D.L.; Blanchard, J.J.;Bellack, A.S.; Nishith, P.; and DeLeon, J. Emotion per-ception and social competence in chronic schizophrenia.Journal of Abnormal Psychology, 105:271-275, 1996.

Nuechterlein, K.H. Vigilance in schizophrenia and relateddisorders. In: Steinhauer, S.R.; Gruzelier, J.H.; and Zubin,J., eds. Handbook of Schizophrenia. Amsterdam, TheNetherlands: Elsevier, 1991. pp. 397^33.

Nuechterlein, K.H., and Dawson, M.E. Information pro-cessing and attentional functioning in the developmentalcourse of schizophrenia disorders. Schizophrenia Bulletin,10:160-203, 1984.

Nuechterlein, K.H.; Dawson, M.E.; Gitlin, M.; Ventura,J.; Goldstein, M.J.; Snyder, K.S.; Yee, CM.; and Mintz, J.Developmental processes in schizophrenic disorders:Longitudinal studies of vulnerability and stress.Schizophrenia Bulletin, 18:387^25, 1992.

Nuechterlein, K.H.; Dawson, M.E.; and Green, M.F.Information-processing abnormalities as neuropsycholog-ical vulnerability indicators for schizophrenia. ActaPsychiatrica Scandinavica, 90(Suppl. 384):71-79, 1994.

Nuechterlein, K.H.; Edell, W.S.; Norris, M.; and Dawson,M.E. Attentional vulnerability indicators, thought disor-der, and negative symptoms. Schizophrenia Bulletin,12:408-426, 1986.

Nunnally, J.M. Psychometric Theory. New York, NY:McGraw-Hill, 1978.

Olney, J.W., and Farber, N.B. Glutamate receptor dys-function and schizophrenia. Archives of GeneralPsychiatry, 52:998-1007, 1995.

Penn, D.L.; Corrigan, P.W.; Bentall, R.P.; Racenstein,J.M.; and Newman, L. Social cognition in schizophrenia.Psychological Bulletin, 121:114-132,1997.

Penn, D.L.; Spaulding, W.D.; Reed, D.; and Sullivan, M.The relationship of social cognition toward behavior inchronic schizophrenia. Schizophrenia Research, 20:327-335, 1996.

Robbins, T.W., and Everitt, B.J. Arousal systems andattention. In: Gazzaniga, M.S., ed. The Cognitive Neuro-sciences. Cambridge, MA: MIT Press, 1995. pp. 703-720.

Schneider, F.; Gur, R.C.; Gur, R.E.; and Shtasel, D.L.Emotional processing in schizophrenia: Neurobehavioralprobes in relation to psychopathology. SchizophreniaResearch, 17:67-75, 1995.

Shakow, D. Segmental set: A theory of the formal psycho-logical deficit in schizophrenia. Archives of GeneralPsychiatry, 6:1-17, 1962.

Silverstein, M.L., and Zerwic, M.J. Clinical psychopatho-logic symptoms in neuropsychologically impaired andintact schizophrenics. Journal of Consulting and ClinicalPsychology, 53:267-268, 1985.

Spaulding, W.D.; Fleming, S.K.; Reed, D.; Sullivan, M.;Storzbach, D.; and Lam, M. Cognitive functioning inschizophrenia: Implications for psychiatric rehabilitation.Schizophrenia Bulletin, 25(2):275-289, 1999.

Spitzer, M. A cognitive neuroscience view of schizo-phrenic thought disorder. Schizophrenia Bulletin,23:29-50, 1997.

Spohn, H.E., and Strauss, M.E. Relation of neurolepticand anticholinergic medication to cognitive functions in

318

Dow




schizophrenia. Journal of Abnormal Psychology,98:367-380, 1989.

Strauss, M.E. Relations of symptoms to cognitive deficitsin schizophrenia. Schizophrenia Bulletin, 19:215-231,1993.

Sweeney, J.A.; Keilp, J.G.; Haas, G.L.; Hill, J.; andWeiden, P.J. Relationships between medication treatmentsand neuropsychological test performance in schizophre-nia. Psychiatry Research, 37:297-308, 1991.

Velligan, D.I.; Mahurin, R.K.; Diamond, P.L.; Hazelton,B.C.; Eckert, S.L.; and Miller, A.L. The functional signifi-cance of symptomatology and cognitive function in schiz-ophrenia. Schizophrenia Research, 25:21-31, 1997.

Wykes, T.; Reeder, C ; Corner, J.; Williams, C ; andEveritt, B. The effects of neurocognitive remediation onexecutive processing in patients with schizophrenia.Schizophrenia Bulletin, 25(2):291-307, 1999.

Young, D.A.; Davila, R.; and Scher, H. Unawareness ofillness and neuropsychological performance in chronic

schizophrenia. Schizophrenia Research, 10:117-124,1993.

Acknowledgment

The authors thank Mary Jane Robertson for her help withthe graphics for this article.

The Authors

Michael Foster Green, Ph.D., is Professor, Department ofPsychiatry and Biobehavioral Sciences, University ofCalifornia Los Angeles, and Health Sciences Specialist,Department of Veterans Affairs VISN 22 Mental IllnessResearch Education and Clinical Center, Los Angeles,CA. Keith H. Nuechterlein, Ph.D., is Professor,Department of Psychiatry and Biobehavioral Sciences,University of California Los Angeles.

319

Dow



Dow



should schizophrenia be treated as a neurocognitive disorder?

Documents