Blood Flow in Monozygotic Twins Discordant and Concordant for Schizophrenia

Regional Cerebral Karen Faith

Berman, MD; E. Fuller Torrey, MD; David G. Daniel, MD; Daniel R.

\s=b\ We addressed several questions regarding hypofunction of the prefrontal cortex ("hypofrontality") in schizophrenia by measuring regional cerebral blood flow during three different cognitive conditions in monozygotic twins who were discordant or concordant for schizophrenia or who were both normal. These questions included the prevalence of hypofrontality, the importance of genetic predisposition, and the role of long-term neuroleptic treatment. Significant differences between affected and unaffected discordant twins were found only during a task linked to the prefrontal cortex, the Wisconsin Card Sorting Test. During this condition, all of the twins with schizophrenia were hypofrontal compared with their unaffected co-twins, suggesting that, if appropriate cognitive conditions and control groups are used, hypofrontality can be demonstrated in the majority of, if not all, patients with schizophrenia. When unaffected cotwins of patients with schizophrenia were compared with twins who were both normal, no differences were observed, suggesting that nongenetic factors are important in the cause of the prefrontal physiologic deficit that appears to characterize schizophrenia. When concordant twins with a high- vs a low-dose lifetime history of neuroleptic treatment were compared, the twin receiving the higher dose was more hyperfrontal in six of eight pairs, suggesting that longterm neuroleptic treatment does not play a major role in

hypofrontality. (Arch Gen Psychiatry. 1992;49:927-934)

of regional cerebral blood flow (rCBF) or metabolism in schizophrenia have been reported by a investigators (see Berman and Weinberger1 for a review). The prefrontal cortex has been an area of partic¬ ular interest. We have previously reported a series of rCBF studies with special focus on the brain region.2"5 Those studies involved patients with chronic schizophrenia dur¬ ing various cognitive tasks, some of which specifically in¬ volved the prefrontal cortex and others that did not. One such task, the Wisconsin Card Sorting Test (WCS),

Abnormal ities number of

Accepted for publication April 23,

1992. From the Clinical Brain Disorders Branch, Intramural Research Program, National Institute of Mental Health, NIMH Neuroscience Center at St Elizabeth's, Washington, DC. Reprint requests to Clinical Brain Disorders Branch, Intramural Research Program, National Institute of Mental Health, NIMH Neuroscience Center at St Elizabeth's, 2700 Martin Luther King Jr Ave, Washington, DC 20032 (Dr Berman).

Downloaded From: by a Thomas Jefferson University User on 02/07/2018

Weinberger,

MD

a neuropsychological test of problem solving and abstract reasoning that requires the use of working mem¬ ory and is known to be a sensitive indicator of prefrontal lobe damage in humans.6,7 In normal control subjects, we have demonstrated, using several techniques, including xenon-133 inhalation rCBF4 and oxygen-15 water positron emission tomography (PET),8 that the neurophysiologic concomitants of performing this test include a significant increase in prefrontal blood flow compared with that dur¬ ing a matched sensorimotor control task. In contrast, in three separate schizophrenia studies, two involving medication-free patients2,4 and one involving patients studied while receiving neuroleptic treatment,3 hypofunction of the prefrontal cortex ("hypofrontality") was dem¬ onstrated during the WCS. However, hypofrontality was not consistently demonstrated in schizophrenic patients during tasks that are not regionally specific or are specific to regions other than the prefrontal cortex.5 The work of other investigators parallels our own; several9,10 have de¬ tected hypofunction of the prefrontal cortex during tasks that may place a physiologic load on this brain area, whereas the results of studies conducted during resting or other nonspecific states have been variable. Taken together, these observations suggest that schizo¬ phrenia is characterized by a cognitively and neurobiologically important dysfunction of the prefrontal cortex that becomes more apparent during physiologic demand. Those studies also raise a number of additional questions that we had the opportunity to investigate by studying pairs of monozygotic twins who were discordant for schizophrenia, concordant for schizophrenia, or both free of psychopathology (ie, normal concordant twins). The first question concerns the distribution of individual data in typical rCBF or PET studies and the prevalence of abnormal prefrontal cortical function in schizophrenia. The design of virtually all published studies has been to compare values for some parameter of frontal blood flow or metabolism between patients and unrelated healthy controls. In many cases, this study design has demon¬ strated that the mean value for the schizophrenic group was significantly lower than that for the group of controls. However, examination of the individual data points typ¬ ically reveals a great deal of overlap between the two groups, with some portion (often a minority) of the patients' values extending beyond the lower end of the normal range (see, for example, Figs 6 and 7 in Weinberger

is

al2, Figs 4 and 5 in Berman et al3, and Figs 1 and 2 in Weinberger et al4). Despite the typical paucity of patient et

values at the upper end of the normal range, which would suggest a shift in the entire distribution toward lower val¬ ues in schizophrenia, one question that has been raised by these observations is whether the difference between the groups might result from just a few low outliers. That is, does hypofrontality affect only a biologically and perhaps etiologically distinct subgroup of the schizophrenic popu¬ lation (eg, patients with prominent "deficit" symptoms) or is it a pathologic process that is characteristic of the disease and that affects, to some degree, most patients who have it? One problem with study designs comparing a patient group with an unrelated control group is that, because there is a great deal of nonspecific variability contributing to the normal range of such data, it is not possible to pre¬ dict what the "normal" value for a given patient should be. Indeed, the concept of a "normal" rCBF value itself is problematic, as it is analogous to asking the question "What is a normal pulse rate?" The answer depends on a number of individual characteristics, including sex, age, height and weight, physical condition, and state of arousal. Many factors are almost impossible to control between groups of unrelated individuals. With no way to ascertain what a patient's value would have been if he or she did not have schizophrenia, it is impossible to know how many patients are "abnormal" compared with their potential functional level. Thus, with unrelated groups, the real prevalence of hypofrontality in schizophrenia cannot be determined and no answer to the question posed above emerges. One approach is to study pairs of monozygotic twins who are discordant for schizophrenia. Because such pairs share an identical genome as well as many socioeco¬ nomic and environmental factors, rCBF measurements for the well co-twin of a monozygotic pair discordant for schizophrenia may be the best indicator of the values that would have characterized the ill twin if he or she did not have schizophrenia. Use of the former as the control to which the latter can be compared offers a means for delin¬ eating the pathophysiologic changes that have occurred in each individual patient and for more closely determining the true frequency of hypofrontality in schizophrenia. A second question raised by the literature has to do with the effects of neuroleptic treatment on frontal lobe func¬ tion. Although the patients in two of our previous studies2,4 had all been medication free for at least 4 weeks at the time of the rCBF study, they, as well as most patients with schizophrenia, had been treated with neuroleptic drugs at some time during the course of their illness. While an ear¬ lier study3 of patients treated with 0.4 mg/kg per day of haloperidol for at least 4 weeks found cognitively related functional deficits in the prefrontal cortex that were virtu¬ ally identical to those found in the medication-free patients (suggesting that neither neuroleptic treatment of relatively short duration nor its withdrawal affected prefrontal blood flow deficits in this illness) few studies have examined the long-term effects of treatment with neuroleptic medication on cerebral blood flow. To address this question, we com¬ pared rCBF between monozygotic co-twins who are con¬ cordant for schizophrenia but who have received different dosages of medications during the course of their illness. Because medication history is one of the few variables dif¬ ferentiating the members of a pair, if long-term adminis¬ tration of neuroleptic medications causes or exacerbates Downloaded From: by a Thomas Jefferson University User on 02/07/2018

hypofrontality, we would expect that the twin who has exposed to more neuroleptic drugs to be the more hypofrontal twin. The third, and perhaps most important, question raised by the previous observations of hypofrontality in schizo¬ phrenia is, assuming that neuroleptic treatment does not play a major role, what is the etiology of such a pathophysiologic dysfunction. The opportunity to study monozygotic twins allowed us to investigate one of the major causative mechanisms that has been proposed in schizophrenia—that of genetic determination. If hypofron¬ tality were an inherited trait that predisposed one to have schizophrenia, we would expect to find hypofrontality in been

the well (unaffected) co-twins of the discordant pairs. Dif¬ ferences in other nongenetic variables, such as premorbid education, nutrition, socioeconomic status, and additional social factors that have been suggested to play a role in the cause of schizophrenia, are also minimized in such a twin

study.

SUBJECTS AND METHODS

Subjects

Twenty-one pairs of monozygotic twins signed informed con¬ sent forms to participate in this study. Monozygosity was estab¬ lished on the basis of history and appearance and then confirmed by testing 19 red blood cell antigens." Diagnostic assessments were made on the basis of a standardized structured diagnostic interview (Structured Clinical Interview for DSM-III-R, parts I and II) and extensive family interviews by an investigator who was unaware of the rCBF results. The patients all fulfilled DSM-III-R criteria for chronic schizophrenia, and all healthy, un¬ affected twins had no evidence of major psychopathologic changes. Ten twin pairs were discordant for schizophrenia (four pairs of women and six pairs of men; mean age, 32 years, age range, 24 to 44 years). At the time of entrance into the study, the minimum length of discordance was 4 years, after which a low rate of conversion to concordance has been reported.12 During an additional 4 years of follow-up, these pairs have remained discordant. Eight pairs were concordant for schizophrenia (one pair of women and six pairs of men; mean age, 32 years; age range, 24 to 41 years). In three pairs (all women; mean age, 28 years; age range, 26 to 32 years), both twins were well (ie, normal concor¬ dant twins). Detailed history of neuroleptic intake was obtained from patient and family interviews and from medical records; milli¬ grams of each neuroleptic drug were converted to fluphenazine equivalents.13 Subjects were screened for major medical problems and for history of significant drug abuse, trauma, and other neu¬ rologic problems. The patients continued to receive their standard doses of neuroleptic medications. Subjects were taking no other medications that might impact rCBF. They were asked to refrain from using nicotine and caffeine before the study. rCBF Measurements cerebral blood flow was measured with the 133Xe inhalation technique, as described by Obrist et al14"16 and mod¬ ified by Deshmuhk and Meyer17 and by Risberg and col¬ leagues.18 Subjects breathed a tracer concentration of ,33Xe gas (185 to 259 µ ^/ of room air) for 1 minute. Thirty-two scintillation detectors were used to monitor the rate of buildup and elimination of 133Xe gamma rays from 32 cortical regions during the 1-minute inhalation and the ensuing 14-minute washout period. Gray-matter rCBF was calculated as the initial slope of each clearance curve. This unitless index measurement has been defined and validated by Obrist and Willdson15 to be a measure of gray-matter blood flow. The end expiratory pC02 was measured continuously with a capnograph. To better ap¬ ply the technique to cognitive activation studies in neuropsy¬ chiatrie patients, we made several modifications.2 Subjects

Regional

Fig 2.—Regional mean (± SEM) gray-matter regional cerebral blood flow values for 10 pairs of monozygotic twins discordant for schizophrenia during three different conditions (black bars indicate unaffected twins; shaded bars, schizophrenic twins). For resting and numbers matching, Wilks' was not significant by multivariate Hotelling's V test; individ¬ ual regions were not significant by univariate analysis. Double asterisk, for the Wisconsin Card

region:

cant. IS

Sorting

Test (WCS), indicates Wilks' X=.105, univariate analyses for WCS prefrontal F=5.48, P=.04. No other univariate comparisons were signifi¬

F=8.5, P=.017;

single asterisk,

indicates initial

slope.

the frontopolar leads of the ternational system.2

rCBF

Fig 1.—Topographic gray-matter group mean regional cerebral blood flow maps showing lateral views of the cortex, with the frontal pole at left, during three different cognitive conditions: resting (top), number matching (center), and the Wisconsin Card Sorting Test (bottom). Color scales are keyed to actual gray-matter initial slope values. studied in the sitting position rather than lying down; a snorkellike mouthpiece and soft nose clamp, rather than the traditional mask, were employed for gas delivery; and we used a method for reproducible placement of the extracranial detectors relative to the orbitomeatal line and the locations of were

Downloaded From: by a Thomas Jefferson University User on 02/07/2018

electroencephalographic

10-20 in¬

Testing Conditions

Each subject underwent three consecutive rCBF measurements separated by a period of at least 30 minutes and performed un¬ der three different testing conditions. The first measurement was made during an eyes-closed "resting" condition, which served to acclimate subjects to the testing procedure. The second and third procedures were conducted during two different cognitive acti¬ vation paradigms that were begun 1 minute before the 133Xe in¬ halation period and performed throughout the ensuing 15-minute rCBF measurement period. The two tasks were presented in counterbalanced sequence to control for the possibility of an or¬ der effect. One task was a semiautomated version of the WCS ; the other was a simple number-matching task designed to serve as a nonspecific sensorimotor baseline activation state, or control task, against which rCBF during the WCS could be compared.24

125r

120

95

-

_l_

-I-

--

Schizophrenic Resting

Unaffected

Schizophrenic Number Matching

Unaffected

Unaffected

Schizophrenic

WCS Test

Fig 3.—Relative distribution analysis (prefrontal indexes) for 10 pairs of monozygotic twins discordant for schizophrenia during three conditions. Asterisk indicates paired t=5.55 and P=.0004 for comparison of affected and unaffected twins during the Wisconsin Card Sorting Test (WCS). Hor¬ izontal bars indicate mean values. Statistical

Analysis

Details and theoretical considerations of our approach to the analysis of these statistically complex data have been presented elsewhere.2 Values for each testing condition were analyzed sep¬ arately. The 32 gray-matter rCBF values for each study were col¬ lapsed into five bilateral cortical regions (prefrontal, precentrai, temporal, parietal, and parieto-occipital), as previously de¬ scribed.2 Because of the relatively small sample sizes compared with the number of variables, right and left hemisphere regions were combined. The five regional values were analyzed on a mainframe computer with the use of Statistical Analysis System packages (SAS82, SAS Institute, Cary, NC). Analyses of the 10 left and right hemisphere regions were also performed; as these yielded no additional information, the results of the nonlateralized five-region analyses are reported herein. Analyses of abso¬ lute blood flow and an index analysis (relative blood flow) were performed to compare the following groups: (1) affected and un¬ affected discordant co-twins, (2) unaffected discordant vs normal concordant twins, and (3) concordant schizophrenic twins with high- vs low-dose neuroleptic exposure. Absolute Regional Analysis.—Absolute rCBF values may be prone to artifacts, such as variations in partition coefficients, pul¬ monary function, and, especially, CO2.19'22 However, as discussed elsewhere,2"5 the effect on rCBF of chronic group differences in Pco2 values that are within the normal range has not been estab¬ lished. Because of these potential méthodologie problems with absolute values, we emphasize the index analysis described be¬ low, which is not affected by pC02. It should be noted, however, that Pco2 values did not differ between co-twins during any of the three testing conditions. Regional gray-matter blood flow was compared between affected and unaffected discordant cotwins with a multivariate analysis (Hotelling's T2 test), an approach that minimizes type I Downloaded From: by a Thomas Jefferson University User on 02/07/2018

This was followed by univariate region-by-region paired comparisons, an approach that minimizes type II error. Hotelling's T2 test is a multivariate statistic that can be applied to repeated measures and/or paired data.23 This statistic evaluates the null hypothesis that the means of the multiple variables (ie, the difference in the five regional values between affected and unaffected twins) do not differ, and in this case that they do not error.

differ from zero. Thus, if the values for affected and unaffected twins were similar for each region, then each difference score would approach zero and Hotelling's T2 test would not be signif¬ icant. Post hoc individual univariate analyses—testing whether each individual regional difference differed from zero—were also

performed.

Relative Distribution (Index) Analysis.—This approach in¬ volves a ratio analysis that considers blood flow to a particular cortical region as a ratio of blood flow to another cortical region. This is a method of quantifying relative rCBF distribution that is probably Pco2-independent. Tlie ratio that has most consistently been shown to differ between schizophrenic and normal control subjects2"4—the prefrontal index—was used. This measure was derived as the ratio of prefrontal rCBF to nonfrontal flow (tem¬ poral, parietal, and parieto-occipital) and was expressed as a per¬ centage. Comparisons between affected and unaffected discor¬ dant twins were conducted with paired f tests. Paired t tests were also used to assess prefrontal indexes for concordant schizo¬ phrenic twin pairs. Within each concordant twin pair, the twin who had received more neuroleptic drugs during his or her life¬ time was compared with his or her co-twin who had received less.

rCBF

Topographic Mapping

To produce qualitative images of group mean rCBF patterns, we used a program developed at the National Institute of Mental Health for topographic mapping. This approach, described else-

No. of

Twin Pair/

-

Age, y/Sex

9 2 4 5 10 10 10 9 6

D2/38/F D3/33/M D4/31/F D5/24/M D6/28/M D7/29/F D8/44/M D9/28/M D10/27/M Mean

C1/31/M C2/38/M C3/24/M C4/25/M C5/41/F C6/31/M C7/28/M C8/27/M Mean

7 7.2

schizophrenic twins

High 0 4 5 1 0 1 1 1 1.6

Nonschizophrenic twins

A 7 2 7

N1/31/F N2/27/F N3/26/F Mean

high neuroleptic dose; low,

low

Affected

Unaffected

Affected

Unaffected

Affected

4 0 2 6 5 4 0 1 3 6 3.1

10 27 20 29 10 9 9 10 15 14 15.3

10 20 22 16 20 12 20 22 25 7 17.4

80 39 62 51 83 82 85 72 55 74 68.3

70 18 36 71 52 70 20 47 43 76 50.3

Low 0 1 4 7 0 1 2 9 3

High 43 18 13 24 54 21 34 29 29.5

Low 45 26 16 9 66 51 45 10 37.2

High 6 50 72 37 8 23 40 20 32

Low 6 23 65 73 9 27 23 80 38

A 12 29 14 18.3

11 17 11 13

A 75 24 62 54

80 54 64 66

8 5 8 7

5.3

indicates

"High

% Conceptual Level of Response

I

Unaffected

Discordant twins D1/38/F

Concordant

% Perseverative Errors

Categories

neuroleptic dose. Nonschizophrenic co-twins were arbitrarily designated A or

B.

during the WCS, a qualitative difference in anterior-to-posterior gradient and in absolute rCBF levels for prefrontal cortex was seen, with lower values in the schizo¬ tions. In contrast,

Twin Pair

*Values

High-Dose

Low-Dose

Co-twin

Co-twin

C1

14 350

8 865

C2 C3

35 027

4 550

39250

21800

C4

28 300

10 840

C5

222 500

155 600

C6

80 500

29 000

C7 C8

101450

400

42 000

31800

Mean

70422

are

fluphenazine equivalents

32 857 in

milligrams.

where,2,24 aids visual

orientation of these complex data. It should be emphasized, however, that the maps were used only for visual orientation and only the actual rCBF values derived from the 32 clearance curves were used in statistical

analysis. RESULTS

Discordant Twins: Affected

vs

Unaffected

Regional Analysis.—Visual orientation maps of initial slope values for cortical gray-matter rCBF during resting and during the two tasks (Fig 1) showed that the patterns were qualita¬ tively similar for affected and unaffected discordant twins dur¬ ing the resting and numbers matching (control task) condiDownloaded From: by a Thomas Jefferson University User on 02/07/2018

phrenic

co-twins.

quantitatively confirmed by the regional analysis (Fig 2). During resting and the numbers matching control task, there were no overall between-group differences revealed by the statistically conservative Hotelling's T2 test (for resting: Wilk's =.25, F=3.0, P=.12; for the numbers matching task: This

was

Wilk's =.44, F=1.3, P=.4), and the more liberal univariate approach disclosed no differences in individual regional val¬ ues. However, for rCBF during the WCS test, the overall Ho¬

telling's

test

was

significant (Wilk's =.10, F=8.5, P

Regional cerebral blood flow in monozygotic twins discordant and concordant for schizophrenia.

We addressed several questions regarding hypofunction of the prefrontal cortex ("hypofrontality") in schizophrenia by measuring regional cerebral bloo...
1MB Sizes 0 Downloads 0 Views