P1: FDR

Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Journal of Abnormal Child Psychology, Vol. 28, No. 1, 2000, pp. 3–14

Genetic Studies of Autism: From the 1970s

into the Millennium

Michael Rutter1

Received July 26, 1999; revision received August 20, 1999; accepted August 20, 1999

Reviewers in the 1960s and early 1970s were skeptical about any substantial role for genetic factors
in the etiology of autism. A realization that the 2% rate of autism in siblings (as estimated at that
time) was far above the general population base rate, and that this suggested a possible high genetic
liability, led to the first small-scale twin study of autism. The replicated evidence from both twin
and family studies undertaken in the 1970s and 1980s indicated both strong genetic influences and
the likelihood that they applied to a phenotype that was much broader than the traditional diagnostic
category of autism. Medical and chromosomal findings also indicated genetic heterogeneity. Advances
in molecular genetics led to genome-wide scans of affected relative pair samples with a positive log of
the odds to base 10 score for a location on chromosome 7. The major remaining research challenges
and the likely clinical benefits that should derive from genetic research are considered in relation to
both current knowledge and that anticipated to emerge from research over the next decade.

KEY WORDS: Autism; quantitative genetics; molecular genetics; functional genomics; clinical practice.

When Kanner (1943) first described the syndrome TWIN STUDIES

of autism, he suggested that it resulted from an inborn
defect of presumably constitutional origin. Nevertheless, An awareness that the logic of these arguments was
over the next 3 decades the possible role of genetic fac- faulty led Susan Folstein and myself (Folstein & Rutter,
tors tended to be dismissed. In part, this was because the 1977a,b) to undertake the first small-scale (N = 21) twin
zeitgeist at that time was one of expecting environmen- study of autism. The reasoning on genetic influences was
tal causes for all forms of psychopathology. This was the false because follow-up studies had shown that it was very
era of supposed “refrigerator” parents of autistic children rare for autistic people to develop love relationships, and
and of “schizophrenogenic” mothers (see Rutter, 1999a). very rare for them to have children, and hence vertical
However, reviews by geneticists were equally dismissive transmission would not be expected; relative to the rate of
(Hanson & Gottesman, 1976). Emphasis tended to be autism in the general population (about 2 to 4 cases per
placed on the lack of vertical transmission (i.e., the rarity 10,000 as defined at that time), the rate of autism was very
with which children with autism had parents with autism), high indeed; and the cytogenetic techniques available in
on the very low rate of autism in siblings (estimated at that the 1960s were quite primitive, so that the failure to show
time at about 2%), and the lack of identified chromosome anomalies was really noncontributory.
anomalies associated with autism (Rutter, 1967). Two main findings from this first twin study were
striking. First, despite the small numbers, there was a
significant monozygotic–dizygotic (MZ-DZ) difference
in concordance. The fact that the population base rate
of autism was so low implied a strong underlying ge-
netic liability. Second, concordance within MZ pairs in-
1 Social, Genetic, and Developmental Psychiatry Research Centre, cluded a range of cognitive and social deficits and not
Institute of Psychiatry, De Crespigny Park, Denmark Hill, London just the seriously handicapping condition of autism itself.
SE5 8AF, UK; e-mail: g.rangel@iop.kcl.ac.uk. This implied that the genetic liability extended beyond

3
0091-0627/00/0200-0003$18.00/0 °
C 2000 Plenum Publishing Corporation
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

4 Rutter

“autism proper.” It also raised questions about the diag-

nostic boundaries of autism and led to an appreciation of
the need to consider the likelihood of a broader phenotype
of autism, or of lesser variants of the same condition.
During the 1970s and 1980s, there was a range of
medical studies of autism demonstrating positive findings
(Rutter, 1999b). A mixed bag of medical conditions were
described in association with autism—most notably at
that time congenital rubella (Chess, 1977; Chess, Korn &
Fernandez, 1971). However, more recently, it became clear
that the association with autism was probably strongest for Fig. 1. Pairwise concordance for autism, social and cognitive
tuberose sclerosis (Smalley, 1998). In addition, there was disorder (based on data from Bailey et al., 1995).
a similarly mixed bag of chromosomal anomalies found
in individuals with autism—with the fragile X anomaly
prominent among them (Gillberg & Wahlström, 1985).
These findings had two important implications. First, it
came to be accepted that systematic screening for medical
conditions and chromosomal anomalies was an essential
part of the diagnostic appraisal of children referred for
possible autism. Second, it was apparent that autism must
be assumed to be etiologically heterogeneous.
During the late 1980s and early 1990s, genetic re-
search into autism advanced through further twin studies,
genetic–family studies, and molecular cytogenetic inves-
tigations. Both a Scandinavian twin study (Steffenburg Fig. 2. Concordance for broader phenotype in MZ and
et al., 1989) and a further British twin study (Bailey et al., DZ pairs discordant for autism (based on data from
1995) confirmed the great strength of genetic influences Le Couteur et al., 1996).
on the underlying liability for autism. Because the latter
study included several crucial methodological improve- volving synergistic interaction among several genes. The
ments over the earlier research, I concentrate on its find- pattern was not compatible with a single-gene Mendelian
ings. There were four key design features: First, there was disorder. Third, the finding that the genetic liability ex-
total population screening for cases, with all clinics and tended to include a broader phenotype was confirmed.
special schools in the country contacted, and all twin regis- Some 90% of monozygotic pairs were concordant for mix-
ters examined. Second, systematic standardized methods tures of social and cognitive deficits qualitatively similar
of diagnosis using both parental interviews (Le Couteur to those found in traditional autism, but milder in degree
et al., 1989) and observation of the child (Lord et al., 1989) (see Fig. 1). This applied, however, to only about 1 in
were employed. Third, there was systematic screening for 10 DZ pairs. Focusing on the 10 MZ and 20 DZ pairs
medical conditions and chromosomal abnormalities in or- discordant for autism or pervasive developmental disor-
der to focus on the study of genetic factors in idiopathic ders (Fig. 2), it was shown that there was a similar con-
autism. Fourth, blood groups were used to test for zygos- trast in concordance for this broader phenotype, the dif-
ity. That was important because the marked behavioral ference being statistically significant (Bailey et al., 1995).
differences associated with autism sometimes led parents A follow-up of the Folstein and Rutter twin sample also
and professionals to infer that the twins were fraternal, showed that this broader phenotype was associated with
whereas in fact they were MZ. important deficits in social functioning that continued into
Four main findings were crucial (see Figs. 1 and 2). adult life (Le Couteur et al., 1996). Fourth, an examination
First, the huge disparity in concordance rates for autism of 16 MZ pairs concordant for autism or atypical autism
between MZ (n = 25) and DZ (n = 20) pairs (60% vs. showed that there was enormous clinical heterogeneity
5%) confirmed the earlier findings on the strength of the even when pairs shared exactly the same genes. Surpris-
genetic influence. Quantitative analyses indicated a her- ingly, individuals within MZ pairs were no more alike in
itability in excess of 90%. Second, the exceedingly low IQ or symptomatology than were pairs of individuals se-
rate of concordance in DZ pairs compared with that in lected at random from different twin pairs (Le Couteur
MZ pairs pointed to the likelihood of epistatic effects in- et al., 1996).
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Genetic Studies of Autism 5

used to estimate the number of genes that likely were to

be involved (Pickles et al., 1995). The findings effectively
ruled out the possibility of just one gene, and suggested
that three or four genes were most probable, but any num-
ber between 2 and 10 genes was a possibility (depending
on the relative strength of effect of any one of these).
The key conclusions from the twin and family stud-
ies up to that point was that the heritability of an under-
lying liability to autism was above 90%, that the liability
extended well beyond the traditional diagnosis, and that
several interacting genes were likely to be involved. So far
so good, but a range of crucial questions remained. These
Fig. 3. Rates of disorder in siblings (based on data
included queries on the role of medical conditions and
from Bolton et al., 1994).
chromosomal anomalies, the clinical indices of genetic
heterogeneity, the defining features of the broader pheno-
FAMILY STUDIES type, the relationship between the broader phenotype and
traditional autism, and the presence and characteristics of
The second line of genetic enquiry was provided by phenocopies.
family studies. These were important to determine the rate
of autism in sibs and in parents, to check family patterns of
transmission in case there were Mendelian variants (this FRAGILE X ANOMALY
cannot be assessed from twin studies), and to better de-
lineate the breadth and pattern of the possible broader The early reports on the fragile X anomaly had led
phenotype. to claims that this was a very common cause of autism
The Maudsley Hospital family study used measure- (Gillberg & Wahlström, 1985). Subsequent studies using
ment methods directly comparable with those in the British the original cell culture methods showed that, to the con-
twin study, and similarly excluded families in which the trary, fragile X anomalies were present in less than 5%
autism was associated with some known medical condi- of individuals with autism (Bailey et al., 1993; Dykens &
tion that was likely to be causal (Fig. 3; Bolton et al., Volkmar, 1997). The development of DNA techniques for
1994). The families of 99 individuals with autism were identification of the fragile X confirmed the infrequency
compared systematically with the 36 families of individu- of fragile X in autism and, most especially, indicated that
als with Down syndrome, using exactly the same methods there were no cases of fragile X in individuals with 1–3%
of measurement. There was direct assessment of all first- of fragile sites on the X chromosome—a proportion that
degree relatives and systematic standardized reports on had been used for positive diagnoses in the early studies
more distant relatives. The rate of autism in the siblings of (Gurling, Bolton, Vincent, Melmer, & Rutter, 1997).
autistic individuals was found to be 3%, with 6% showing We need to pause, therefore, to consider what the
some form of pervasive developmental disorder (PDD). flaws were in the original fragile X claims. Three features
No cases of autism or PDD were found in the siblings of were probably most important: (1) a reliance on dubious
individuals with Down syndrome. As in the twin study, a cell culture findings giving rise to only 1–3% of fragile
broader phenotype of autism, comprising mixed patterns sites; (2) a lack of care in clinical diagnosis; and (3) the
of cognitive and social deficits and repetitive stereotyped fallacious assumption that positive findings from small
interest patterns, were even more frequent. Depending samples are more likely to be valid (see Pocock, 1983).
on how stringent a definition was used, the comparative
rates of the broader phenotype, as compared with Down
syndrome families, were about 12% versus 2% or 20% CLINICAL INDICES OF GENETIC
versus 3%. HETEROGENEITY
The relative increase in risk in siblings was some-
where between 30-fold and 100-fold, depending on the It might be expected that etiological heterogeneity
assumptions made about the base rate in the general pop- would be associated with differences in clinical phenom-
ulation. The increase in risk in MZ co-twins was 10 times ena. It is possible that that will prove to be the case but,
that. The falloff in rate from MZ to DZ twins, together up to now, the indications are weak. The family study
with that from first-degree to second-degree relatives, was findings had shown that the more severe the autism, the
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

6 Rutter

greater the familial loading tended to be, as indexed by ei- lems (Landa, Wzorek, Piven, Folstein, & Isaacs, 1991;
ther symptomatology or verbal IQ, but this seemed to be a Landa et al., 1992), of social abnormalities (Piven et al.,
graded phenomenon rather than a categorical distinction. 1990, 1991), and of unusual personality features (Piven
Some data on families (August, Stewart, & Tsai, 1981) had et al., 1994). The early findings had particularly empha-
suggested that autism that was accompanied by profound sized the role of language delay but, although this does
mental retardation might be somewhat different from the seem to be part of the overall picture in some cases, the
rest of autism in which the nonverbal IQs were above 50. later findings have suggested the probably greater im-
Accordingly, the Maudsley Hospital group undertook a portance of social deficits (Bailey, Palferman, Heavey, &
further family study to determine whether this might be Le Couteur, 1998; Folstein & Piven, 1991; Piven, Palmer,
the case (Starr et al., 1999; Pickles et al., in press). The Jacobi, Childress, & Arndt, 1997; Rutter et al., 1997.)
findings were largely negative. The rate of autism and of As the evidence accumulated from a range of stud-
the broader phenotype in the relatives was not significantly ies of varying quality, more and more clinical features,
different from the rates in the first family study. The only including affective disorder and social anxiety (DeLong
possible lead was the uncertain indication that cognitive & Nohria, 1994; Smalley, McCracken, & Tanguay, 1995),
problems in the relatives might be more common when the came to be added to possible variations of the broader
autism was associated with profound retardation. What did phenotype. Clearly, there was a danger that it would be-
emerge, however, from the combination of the two studies come unhelpfully overinclusive and it was important to
was the finding that the linear association between sever- have a means of ruling out, as well as ruling in, possi-
ity of autism and familial loading seemed to apply only bilities. In that connection, two largely negative findings
to cases of autism in which there was some useful speech (with respect to cognitive impairment and to depression)
(Pickles et al., in press). The implication is that, when warrant mention. First, Fombonne, Bolton, Prior, Jordan,
autism is associated with a very severe lack of language & Rutter’s (1997) detailed analysis of cognitive patterns
skills, it might be genetically different in some way. in relatives in the Maudsley Hospital family study showed
that neither low IQ nor specific problems in reading or
spelling showed an increased loading in the families of
PHENOCOPIES individuals with autism if these problems were not ac-
companied by other manifestations of the broader phe-
Over the past 10 years, evidence has accumulated, notype. Somewhat contrary to earlier impressions, it now
however, on the existence of what seemed to be pheno- seems that such specific cognitive deficits, when they oc-
copies—meaning clinical pictures that look like autism but cur in isolation, are not indicators of a genetic liability to
that are not due to the same genetic liability. Thus, atypical autism. On the other hand, language impairments do con-
syndromes of autism have been found to be associated with stitute an important part of the broader phenotype; it is just
congenital blindness (Brown, Hobson, & Lee, 1997), with that they do not ordinarily appear to be related to autism if
profound institutional privation (Rutter, Andersen-Wood, they occur without social deficits or circumscribed inter-
et al., 1999), and with a mixed bag of medical conditions or est patterns. The same findings also showed the apparently
with profound mental retardation (Rutter, Bailey, Bolton, surprising finding that the relatives tended to show a cog-
& Le Couteur, 1994). The evidence to date does not show nitive pattern with verbal skills that were superior to visual
decisively that these syndromes do not involve any of the or spatial skills—namely, the opposite of what is ordinarily
same genetic liability but that is the implication of the found in autism. Piven and Palmer (1997) found the same.
findings. As in the Johns Hopkins study (Piven et al., 1994),
the Maudsley Hospital study showed an increase in the
rate of unusual personality traits in the relatives of autistic
BROADER PHENOTYPE individuals (Murphy et al., 1999). The traits of shyness
and aloofness were especially frequent in those with other
During the past 20 years, there has been a grow- manifestations of the broader phenotype and the traits of
ing number of family studies of one kind or another that anxiety and oversensitivity were especially frequent in
have sought to delineate more precisely the nature of relatives who also had anxiety or depressive disorders.
the broader phenotype of autism, together with identifi- The rates of clinically significant affective disorder were
cation of its boundaries. The Johns Hopkins study, ini- also increased in the relatives of individuals with autism
tially planned in collaboration with the Maudsley Hospital (Bolton, Pickles, Murphy, & Rutter, 1998).
study, was particularly important because of its evidence The finding that a phenomenon is increased in rela-
on the probable importance of pragmatic language prob- tives does not, in itself, necessarily mean that it reflects a
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Genetic Studies of Autism 7

genetic liability. It is important to go on to test alternative availability of good, well-standardized diagnostic mea-
possibilities. This could be done, indirectly, in the family sures of proven reliability. The disadvantages lay in the
studies by determining whether depression was associ- absence of good candidate genes, the lack of evidence on
ated at the individual level with the broader phenotype of the mode of genetic transmission, and considerable un-
cognitive and social deficits and similarly whether it was certainty on the boundaries of the phenotype. This meant
associated at the family level. The findings were clear-cut that linkage studies of large families were inappropriate,
in indicating that it was not at either level. Depression was the lack of candidate genes meant that there were poor
associated with depression in other family members but leads for any association study and hence the choice of an
not with the broader phenotype. At an individual level, too, affected sib-pair design was the obvious way to go.
it overlapped with other manifestations of affective disor- It was possible to use well-standardized diagnostic
der but not with cognitive or social deficits. Also, unlike measures to focus on a well-validated narrow diagnos-
the broader phenotype, the familial loading for affective tic concept. It was appropriate, and possible, to exclude
disorder was not associated with the severity of autism. individuals with profound retardation, to rule out cases
The cause of the increased rate of depression in the fami- that might be due to medical causes, and, in addition, in
lies of individuals with autism remains unclear but it does view of the evidence on phenocopies, it seemed sensible to
not seem to reflect a genetic liability to autism. exclude individuals who had experienced unusually pro-
Putting together the findings on the broader pheno- found psychological privation. Also, systematic screening
type, it can be concluded that there is good evidence from for multiplex families (meaning families with at least two
both twin and family studies that the genetic liability in- affected members) was essential.
cludes patterns of social and cognitive deficits accompa- The likelihood of genetic heterogeneity, together with
nied by circumscribed interest patterns that are very sim- the fact that the search needed to be for multiple suscep-
ilar to autism in quality but much milder. However, there tibility genes, meant that a large sample of sib pairs was
seem to be two crucial differences between the broader essential. Accordingly, an international consortium was
phenotype and autism as traditionally discussed. Unlike established, standard methods were set up to check on
autism, the broader phenotype is not associated with men- reliability, and a total genome scan was undertaken by
tal retardation and it is not associated with epilepsy. As Monaco and colleagues at the Wellcome Trust Centre for
yet, we do not know why that is so. Questions arise as to Human Genetics. The findings produced a positive log of
whether the broader phenotype represents a lesser “dose” the odds to base 10 (LOD) score for a location on chro-
of genetic liability, a different pattern of susceptibility mosome 7, together with more weakly positive findings
genes, or some kind of “two-hit” mechanism, in which for locations on five other chromosomes (International
some additional risk factor is required in order to take in- Molecular Genetic Study of Autism Consortium, 1998).
dividuals over the threshold from the broader phenotype The LOD score is a statistical term that quantifies the like-
to a more seriously handicapping disorder. lihood that two loci are linked or unlinked (Burmeister,
1999), in this case a linkage between the location on chro-
mosome 7 and one of the postulated loci for susceptibility
MOLECULAR GENETICS genes for autism. A score of +3 or more is commonly
accepted as showing linkage; in the study, the score found
During the 1990s, it became evident that use of mole- for the UK sample exceeded that cutoff. This finding con-
cular genetic strategies was indicated and that the technol- stituted the first crucial step in the search for susceptibility
ogy had reached a point at which when the enterprise was genes for autism—meaning genes that are causally impli-
feasible. In particular, the development of robotic tech- cated in the liability to autism but which do not, on their
niques and the availability of a very large number of poly- own, cause it directly. They provided the basis for a pos-
morphic microsatellite genetic markers had made a to- sible breakthrough but it is important to recognize that it
tal scan of the genome a practical possibility (Maestrini, was the beginning of the beginning, not the end of the
Marlow, Weeks, & Monaco, 1998). In planning a molec- beginning, and certainly not the end of the search.
ular genetic study, there were three features that made The history of psychiatric genetics is full of early
autism a very good disorder for this approach and there claims for positive findings that subsequently could not
were three features that presented problems. The pluses be replicated and have had to be withdrawn, or at least
lay in the very consistent evidence of a strong genetic very substantially modified (Rutter, 1994; Rutter, Silberg,
liability, with a heritability for the underlying liability to O’Connor, & Simonoff, 1999a,b). Several further imme-
autism above 90%, the evidence of a relatively small num- diate steps needed to be taken. To begin, the International
ber (probably less than 10) of susceptibility genes, and the Consortium increased the size of the affected-relatives
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

8 Rutter

sample to determine whether the LOD score holds up; 1997, 1998) from the TDT. Genes expressed in the brain
it does. Even more crucially, as is the rule throughout sci- (i.e., affecting brain structure and/or function) can be sys-
ence, replication in independent samples is mandatory. No tematically tested for autism susceptibility variants, with
finding is valid until it has been confirmed by other investi- putative variants tested in patients and controls.
gators. Discussions with other research groups throughout Identification of a susceptibility gene on chromo-
the world who have comparable sib-pair samples has led some 7 will, of course, only deal with a small part of
to informal reports that they have similar, albeit weaker, the genetic liability and it will be necessary to go through
positive findings in much the same position on chromo- similar search processes, using larger samples, to identify
some 7 (see also Philippe et al., 1999). What is needed other susceptibility genes. However, the power to detect
now is some form of meta-analysis across the different them will have been increased by the identification of the
studies and that is being planned. gene on chromosome 7.
One of the limitations of linkage studies (examining Association strategies have the advantage over link-
co-inheritance, that is, inheritance within families in which age strategies of being better able to detect very small
there is linkage between the gene locus being studied and genetic effects (Risch & Merikangas, 1996). Up to now,
the condition being investigated) is that a positive finding however, they have relied on the availability of candidate
covers a large area on the chromosome. Association stud- genes. These have been singularly lacking in relation to
ies (based on the quite different strategy of using linkage autism. Nevertheless, some groups have attempted this
disequilibrium to search for differences between cases and approach, but the findings so far (with the possible excep-
controls in allelic patterns) can help to narrow this, and, tion of the chromosome 15 finding—Cook et al., 1998;
in any case, it is always desirable to check whether a find- but see Maestrini et al., 1999) have been contradictory
ing holds up using different strategies. The transmission and inconclusive (Cook et al., 1997; Klauck, Poustka,
disequilibrium test (TDT; Spielman & Ewens, 1996) con- Benner, Lesch, & Poustka, 1997). The development of
stitutes the best method because it controls for stratifica- the technique of DNA pooling (i.e., combining DNA sam-
tion biases and because it tests for both association and ples across cases and similarly across controls) provides
linkage (Malhotra & Goldman, 1999). The stratification a possible way forward (Barcellos et al., 1997; Daniels
bias arises because cases and controls may differ in their et al., 1998). It should be possible to use this technique
allelic patterns because they differ in their ethnic origins in a way that controls for stratification biases (meaning
rather than for any reason to do with the disorder being that cases and controls are genetically different for rea-
studied. In that connection, it is important that the crucial sons that have nothing to do with autism). The use of
ethnic variations may be quite subtle, such as the differ- parents as the control strategy, as in the TDT, overcomes
ence between having a Finnish rather than Swedish back- the problem. Accordingly, at least in theory, it should be
ground. The TDT has the advantage of requiring DNA possible to undertake genome scans, using DNA pool-
samples from only the affected child and both parents. ing and an association strategy. This has yet to be shown
Accordingly, it can and should be undertaken with single- to be a successful strategy and there are two main issues to
ton samples as well as multiplex family samples. That is be dealt with. First, the number of markers required will
under way in the laboratories of several of the research be very much greater than in linkage studies because the
groups in the international consortium. In addition, it is association strategies can produce positive findings only
desirable, too, to make use of quantitative trait loci (QTL) in relation to a susceptibility gene that is very close to the
approaches predicated on the expectation that susceptibil- markers used or is the trait marker itself. Second, there
ity genes may have effects on gradations in a continuously are statistical problems in determining the significance of
distributed phenotype, and not just on the extreme disor- case-control differences when a very large number of gene
der (Allison, 1997; Fulker & Cherny, 1996; Kruglyak & markers have to be tested. Nevertheless, this does consti-
Lander, 1995). tute one additional possible way forward.
Genomic clone contigs for chromosome 7q will need In addition, there needs to be a follow-through on
to be identified from genome maps or library screening, findings that apply to specific genetic disorders associated
and then sequenced. These are cloned short sequences with autism. Tuberous sclerosis constitutes one possible
of human DNA that have been inserted into some other lead because of the evidence that a significant minority
organism. Then, computer analyses can be used to re- (less than 5%) of autistic individuals have tuberous scle-
assemble the shorter sequences into longer ones (Watson, rosis and because the rate of autism has been shown to
Gilman, Witkowski, & Zoller, 1992). Analyses of can- be much increased in individuals with tuberous sclero-
didate genes can focus on those at the already identified sis (Smalley, 1998). Chromosome 15 anomalies have also
peak of the linkage disequilibrium curve (Merriman et al., provided a lead that needs to be followed (Cook et al.,
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Genetic Studies of Autism 9

1997; Schroer et al., 1998). It is notable that the associa- serotonin level, and other correlates of autism might be
tion between autism and an intrachromosomal duplication informative.
of 15q11-q13 seems to depend on maternal transmission. One obvious priority concerns research to under-
The possibility of genomic imprinting needs to be con- stand the phenomenon of the broader phenotype. Up to
sidered. The precise mechanism involved in imprinting now, it has largely been conceptualized and measured as
remains uncertain but the term means that the expression a lesser variant of autism. Its assessment needs to rely on
(or effect) of a gene is influenced by whether it is inherited a combination of social deficits, communicative deficits,
from the father or the mother. It seems that, in some way, and repetitive stereotyped interests as observed and as re-
the gene is marked (imprinted) in the process of trans- ported by the individuals themselves and by others. As
mission from parent to offspring. Several such genes have always, multiple methods of assessment, using different
been identified in mice and humans (Barlow, 1995) and data sources, are necessary. Some guidance may be pro-
one is known to cause both Angelman and Prader-Willi vided by the age of first manifestation. In addition, there is
syndromes (both are neuropsychiatric conditions) accord- considerable potential in using the pattern of social cog-
ing to whether the gene comes from the mother or the nitive abnormalities as a validating criterion. There are
father (see Rutter et al., 1999a). a variety of indications in the literature that this might
It may be anticipated confidently that, over the next be fruitful (see, e.g., Hughes, Leboyer, & Bouvard, 1997;
few years, these various research approaches will be suc- Hughes, Plumet, & Leboyer, 1999) but, to apply the leads
cessful in identifying several susceptibility genes for in a systematic way, such abnormalities would need to be
autism. However, in itself, that will not have clinical bene- dimensionalized and quantified, and then related to other
fits. Much more needs to be done (Wahlsten, 1999). Thus, measures of social functioning. Probably the greatest po-
it is necessary to go on to clone the susceptibility genes, to tential lies in some combination of theory of mind skills as
determine the gene effects on proteins, and to delineate the tested in naturalistic ways (see, e.g., Baron-Cohen, 1995;
route from these gene effects to the autism phenotype. The Happé, 1994; Heavey, Phillips, Baron-Cohen, & Rutter, in
application of this approach to traditional autism clearly press) and pragmatic qualities in conversational language
constitutes the first application to be undertaken. Never- (Landa et al., 1991, 1992), but central coherence features
theless, it will be necessary to go on to validate the broader (Frith & Happé, 1994) and aspects of executive plan-
phenotype and to apply the range of molecular genetic ning (Hughes, et al., 1999; Ozonoff, Rogers, Farnham,
strategies to that phenotype. It will be important to seek & Pennington, 1993; Piven & Palmer, 1997) also need to
to develop dimensional phenotype measures and, having be considered. However, it is likely that executive plan-
done so, to use QTL approaches with markedly discordant ning deficits may well prove to lack diagnostic specificity
sib pairs. At present, we do not know whether there is a (Griffith, Pennington, Wehner, & Rogers, 1999).
dimensional equivalent of autism but that remains a real Two key comparisons are required. First, there is a
possibility that needs to be explored. need to determine the difference in frequency of the lesser
variant between relatives of probands with autism and rel-
atives of probands with schizophrenia or some other con-
NEED FOR QUANTITATIVE GENETIC STUDIES dition that is genetically distinct from autism but which
involves a familial loading for social problems. This needs
It is sometimes thought that the day of quantitative to be followed by an analysis to determine the differences
genetics is over, but it is not. Thus, in relation to autism, in the particular pattern of the phenotype between those
there is a need for twin and family studies of broader phe- with abnormalities in the two groups. The point is that
notypes. At the moment, we know something about the there are many different causes of social communicative
occurrence of such phenotypes in the relatives of indi- and behavioral problems in human beings and it is cru-
viduals with autism but there are no adequate studies in cial to be able to determine what is distinctively different
which the starting point is probands with the broader phe- about those that index autism.
notype. There is also more scope for studies of within- How can we get to that point? The start, I suggest, is
and across-MZ pair variations in order to explore possible a focus on abnormalities found in individuals with autism
clinical indicators of genetic heterogeneity. For example, who have normal levels of nonverbal intelligence. The first
this needs to be done with respect to both epilepsy and task, then, is to determine the confluence among the varied
language level. The cause of the marked male preponder- measures; that is, the extent to which, say, theory of mind
ance in cases of autism is quite unknown and leads might deficits is associated with abnormalities in social reci-
be provided by male/female comparisons within multi- procity and pragmatic language impairment. This needs to
plex families. Also, family studies of head circumference, be followed by the validation of group differentiation and
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

10 Rutter

to go on to the establishment of means of diagnosis at an

individual level. Finally, the diagnosis of the broader phe-
notype must be related to studies of brain functioning (as
by functional imaging) and to genetics (using molecular
techniques). As already noted, there are important dif-
ferences between the broader phenotype and traditional
autism, and it is necessary to undertake research to deter-
mine whether this is a consequence of a different genetic
“dose,” different gene patterns, or the operation of some
nongenetic factor, as yet to be identified.
In the past, psychiatric geneticists have tended to
think of genetic effects as if they operated at a syndrome
or diagnostic level. Of course, that may be how they work.
On the other hand, it is crucial to consider the possibility Fig. 4. The long journey from genes to therapy.
that the genetic effects operate on components of autism,
rather than the syndrome as such. For example, there are bringing together quite diverse areas of science. As is
indications that that could be the case in relation to read- well recognized, susceptibility gene identification has be-
ing difficulties (Fisher et al., 1999; Gayán et al., 1999; come possible through advances in molecular genetics and
Grigorenko et al., 1997). This means that genetic stud- through the information provided by sequencing the hu-
ies must be undertaken in relation to both possibilities. man genome. The development of single nucleotide poly-
That would require the separation and dimensionalization morphisms and of microarray chips will further aid this
of different components (not just in relation to symptom endeavor (Brown & Hartwell, 1998; Lander, 1999; Watson
patterns but also in terms of cognitive level and unusual & Akil, 1999). The steps required to understand the func-
circumscribed cognitive talents). In addition, it will be tion of these genes are many and varied and it has to be
important to consider the possibility that the susceptibil- recognized that there is not yet a well-established path for
ity genes for autism may include some that are involved in this research process. On the one hand, there is the research
the liability to developmental language disorders, with the focusing directly on the genes themselves with various an-
genetic-effects ones that are not diagnosis-specific. In that imal models a crucial part of the process, but with bioin-
connection, it is interesting that the locus found for an un- formatics essential, too, in the linking up of information
usual form of familial speech/language disorder (Fisher, across species and across different research approaches.
Vargha-Khadem, Watkins, Monaco, & Pembrey, 1998) is Transgenics, gene insertion, and gene “knockout” tech-
situated quite close to that on chromosome 7, which seems niques have a crucial part to play in the identification of
to link with autism. Recent research has shown that the gene function because they provide an experimental ap-
“theory of mind” deficits associated with autism are also proach to gene actions (Capecchi, 1994; Crabbe, Belknap,
found with some severe developmental disorders of re- & Buck, 1994; Sibilia, & Wagner, 1996). However, these
ceptive language (Clegg, Hollis, & Rutter, 1999). Also, it methods are likely to be difficult to apply in autism because
will be important to consider the possibility that the risk of the problems of measuring the relevant phenotypic fea-
processes involve the absence of protective genes as well tures in mice (Rutter et al., 1999a). Also, it will be cru-
as the presence of susceptibility genes. cial to use the new technologies of proteomics (the study
of protein properties) and transcriptomics (the variety of
technologies for screening for messenger RNA content).
FUNCTIONAL GENOMICS All of this constitutes components of the broader multidis-
ciplinary field of functional genomics, a field that is still
In looking ahead into the Millennium, we need to at quite an early stage of development.
consider what is going to be required for the genetic find- Having found out the nature of gene functions, the
ings in autism to be translated into clinically useful means next step involves identification of the risk process by
of prevention or treatment. Genetic evangelists tend to im- which the anomalies of gene function lead to the phe-
ply that this will automatically follow in a relatively short notype of autism. This will require research to understand
period of time. It will not. There is a long journey to be un- the workings of the biological system (molecular cell bi-
dertaken (Fig. 4) to go from the identification of genes to ology) but it will also involve the determination of how
the development of effective new therapies (Chakravarti, the genetic risks interact with developmental processes
1999). Also, it is evident that this journey will involve or with environmental hazards of some kind (requiring
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Genetic Studies of Autism 11

molecular epidemiological research). If all of this is suc- to be a focus on possible protective genes as much as on
cessful in its aims, it then should be possible to move on to nongenetic influences. This focus might throw light on
the further step of outlining the disease model. That will why autism is so much more frequent in males, as well
require integrative biological studies that are combined as on the mechanisms involved in the transition from the
with experimental testing in order to determine whether broader phenotype to the handicapping disorder of autism.
the model is correct in its details as well as in its overall Sixth, there is the potential of molecular genetic findings
form. Of course, even the construction of a valid disease for leads on effective drug treatments. This will involve a
model does not in itself provide clinically effective meth- focus on specific pharmacological actions and, most espe-
ods of prevention or treatment. That requires yet further cially, on individual differences in response to medication.
research steps, which there is not space to consider here. The extent to which this will lead to effective interventions
will depend to a considerable extent on what proves to be
the nature of the underlying neural processes and on how
CLINICAL BENEFITS OF GENETIC RESEARCH they predispose to autism.
Finally, there is the potential of molecular genetic
Let me end, nevertheless, on a more positive note findings in aiding the identification of environmental risks.
by outlining the potential benefits that may be anticipated This will require a focus on nature–nurture interplay and
from these research endeavors. The topic may be con- on the delineation of the role of environmental media-
sidered under the broad question of how genetic findings tion of risks. Autism is a complex multifactorial disor-
might matter for families that contain an autistic individ- der and the evidence suggests that the risks are likely to
ual. There are seven main benefits, some of which have involve nongenetic, as well as genetic, factors, although
already been obtained. First, the genetic findings have al- the latter are probably much the most powerful influence.
ready had an effect on the prevailing concepts of the na- At present, we know very little about these nongenetic
ture of autism. We now focus on genetically influenced factors and it may be that the nongenetic vulnerabilities
neurodevelopmental deficits rather than on maladaptive could be proven to reside in random developmental vari-
patterns of upbringing. There are, of course, still queries ations (Goodman, 1991; Kurnit, Layton, & Matthysse,
on the nature of these deficits, and the possibility of inter- 1987; Molenaar, Boomsma & Dolan, 1993) rather than ei-
acting environmental risks has not been ruled out. ther obstetric or postnatal influences (Bolton et al., 1997).
Second, the findings raise the need for the availabil- Still, we need to understand what these factors are and
ity of genetic counseling. In such counseling, it is nec- how they operate. It is possible, in that connection, that
essary to focus on the contrast between the rather low dermatoglyphic patterns (Davis & Bracha, 1996) or mi-
level of absolute risk and the high level of relative risk of nor congenital anomalies (Meyers, Elias, & Arrabal, 1995)
autism in other family members. The queries in this con- might provide some index of developmental perturbations.
nection mainly concern the risks for autism-related but
less-handicapping disorders.
Third, there is the potential of molecular genetic find- CONCLUSION
ings to provide a better identification of the broader phe-
notype. This will allow a focus on which developmental The past 30 years has seen tremendous advances
delays, cognitive features, and social deficits are part of an in the understanding of the role of genetic influences in
autism liability. However, the connections between autism autism. We have come a long way from the views of the
and the broader phenotype may prove to be quite complex early 1970s when even behavior geneticists doubted that
and difficult to determine. there were any genetic effects to be investigated in autism.
Fourth, there is the potential of molecular genetic We now know that, despite that early skepticism, genetic
findings for leads on biological research that will identify influences are hugely important in the liability to autism,
the causal neural processes that underlie the development that they involve the operation of several interacting genes,
of autism. These will allow a focus on the specific neuro- and that the liability extends well beyond traditional con-
biological effects of susceptibility genes and on the routes cepts of a handicapping disorder usually accompanied by
by which such effects predispose to autism. As I have in- some degree of mental retardation, and often with the de-
dicated, the identification of causal processes could lead velopment of epilepsy in adolescence. In concluding this
to effective means of prevention or intervention, but the paper, I have tried to look ahead into the next 30 years.
research route is likely to prove long and arduous. I think there is every reason to be optimistic about the
Fifth, there is the potential of molecular genetic find- likelihood that the genetic research will lead to biological
ings for leads on possible protective factors. There needs studies that will, at last, delineate the underlying causal
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

12 Rutter

neural processes and that their understanding will have Bolton, P., Murphy, M., Macdonald, H., Whitlock, B., Pickles, A., &
substantial clinical benefits. For this research program to Rutter, M. (1997). Obstetric complications in autism: Consequences
or causes of the condition? Journal of the American Academy of
be successful, it will need to combine clinical, epidemi- Child and Adolescent Psychiatry, 36, 272–281.
ological, and basic science strategies and it will depend Bolton, P., Pickles, A., Murphy, M., & Rutter, M. (1998). Autistic, affec-
on crucial advances in several of these fields. Only time tive and other psychiatric disorders: Patterns of familial aggregation.
Psychological Medicine, 28, 385–395.
will tell whether my optimism on the likely success of Brown, P. O., & Hartwell, L. (1998). Genomics and human disease—
this endeavor will prove to be justified. If it is successful, variations on variation. Nature Genetics, 18, 90–93.
however, clinical practice with individuals suffering from Brown, R., Hobson, R. P., & Lee, A. (1997). Are there autistic-like
features in congenitally blind children? Journal of Child Psychology
autism in its several varieties is likely to be transformed. and Psychiatry, 38, 693–703.
Burmeister, M. (1999). Basic concepts in the study of diseases with
complex genetics. Biological Psychiatry, 45, 522–532.
Capecchi, M. R. (1994). Targeted gene replacement. Scientific American,
ACKNOWEDGMENTS 270, 34–41.
Chakravarti, A. (1999). Population genetics—making sense out of se-
quence. Nature Genetics, 21, 56–60.
This paper was given as the Presidential Address at Chess, S. (1977). Follow-up report on autism in congenital rubella. Jour-
the Ninth Biennial International Society for Research in nal of Autism and Childhood Schizophrenia, 7, 69–81.
Chess, S., Korn, S. J., & Fernandez, P. B. (1971). Psychiatric disorders
Child and Adolescent Psychopathology Conference held of children with congenital rubella. New York: Brunner/Mazel.
in Barcelona, Spain, July 16–20, 1999. The ideas and find- Clegg, J., Hollis, C., & Rutter, M. (1999, June). Developmental language
ings in this paper owe much to the many splendid collab- disorders: A longitudinal study of cognitive, social, and psychiatric
functioning. Paper presented at the biannual meeting of the Interna-
orators with whom it has been my good fortune to work tional Society for the Study of Child and Adolescent Psychopathol-
in the field of the genetics of autism. I would like to ex- ogy, Barcelona, Spain.
press particular appreciation of the contributions over the Cook, E. H., Courchesne, R. Y., Cox, N. J., Lord, C., Gonen, D., Guter,
S. J., Lincoln, A., Nix, K., Haas, R., Leventhal, B. L., & Courch-
years of Anthony Bailey, Patrick Bolton, Susan Folstein, esne, E. (1998). Linkage-disequilibrium mapping of autistic disor-
Eric Fombonne, Ann Le Couteur, Catherine Lord, An- der, with 15q11-13 markers. American Journal of Human Genetics,
thony Monaco, and Andrew Pickles. To an even greater 62, 1037–1083.
Cook, E. H. J., Lindgren, V., Leventhal, B. L., Courchesne, R., Lincoln,
extent, I am indebted to the numerous families who have A., Shulman, C., Lord, C., & Courchesne, E. (1997). Autism or
participated in our studies. atypical autism in maternally but not paternally derived proximal
15q duplication. American Journal of Human Genetics, 60, 928–
934.
Crabbe, J. C., Belknap, J. K., & Buck, K. J. (1994). Genetic animal
REFERENCES models of alcohol and drug abuse. Science, 264, 1715–1723.
Daniels, J., Holmans, J., Williams, N., Turic, D., McGuffin, P., Plomin,
R., & Owen, M. J. (1998). A simple method for analyzing mi-
Allison, D. B. (1997). Transmission-disequilibrium tests for quantitative crosatellite allele image patterns generated from DNA pools and
traits. American Journal of Human Genetics, 60, 676–690. its application to allelic association studies. American Journal of
August, G. J., Stewart, M. A., & Tsai, L. (1981). The incidence of cogni- Human Genetics, 62, 1189–1197.
tive disabilities in the siblings of autistic children. British Journal Davis, J. O., & Bracha, H. S. (1996). Prenatal growth markers in
of Psychiatry, 138, 416–422. schizophrenia: A monozygotic co-twin control study. American
Bailey, A., Bolton, P., Butler, L., Le Couteur, A., Murphy, M., Scott, S., Journal of Psychiatry, 153, 1166–1172.
Webb, T., & Rutter, M. (1993). Prevalence of the fragile X anomaly DeLong, R., & Nohria, C. (1994). Psychiatric family history and neu-
amongst autistic twins and singletons. Journal of Child Psychology rological disease in autistic spectrum disorders. Developmental
and Psychiatry, 34, 673–688. Medicine and Child Neurology, 36, 441–448.
Bailey, A., Le Couteur, A., Gottesman, I., Bolton, P., Simonoff, E., Yuzda, Dykens, E. M., & Volkmar, F. (1997). Medical conditions associated
F. Y., & Rutter, M. (1995). Autism as a strongly genetic disorder: with autism. In D. J. Cohen & F. R. Volkmar (Eds.), Handbook of
Evidence from a British twin study. Psychological Medicine, 25, autism and pervasive developmental disorders (pp. 388–410). New
63–77. York: Wiley.
Bailey, A., Palferman, S., Heavey, L., & Le Couteur, A. (1998). Autism: Fisher, S. E., Vargha-Khadem, F., Watkins, K. E., Monaco, A. P., &
The phenotype in relatives. Journal of Autism and Developmental Pembrey, M. E. (1998). Localisation of a gene implicated in a severe
Disorders, 28, 369–392. speech and language disorder. Nature Genetics, 18, 168–170.
Barcellos, L. F., Klitz, W., Field, L. L., Tobias, R., Bowcock, A. M., Fisher, S. E., Marlow, A. J., Lamb, J., Maestrini, E., Williams, D. F.,
Wilson, R., Nelson, M. P., Nagatomi, J., & Thomson, G. (1997). Richardson, A. J., Weeks, D. E., Stein, J. F., & Monaco, A. P.
Association mapping of disease loci, by use of a pooled DNA ge- (1999). A quantitative-trait locus on chromosome 6p influences
nomic screen. American Journal of Human Genetics, 61, 734–747. different aspects of developmental dyslexia. American Journal of
Barlow, D. P. (1995). Gametic imprinting in mammals. Science, 270, Human Genetics, 64, 146–156.
1610–1613. Folstein, S. E., & Piven, J. (1991). Etiology of autism: Genetic influences.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory Pediatrics, 87, 767–773.
of mind. Cambridge, MA: MIT Press. Folstein, S., & Rutter, M. (1977a). Genetic influences and infantile
Bolton, P., Macdonald, H., Pickles, A., Rios, P., Goode, S., Crowson, autism. Nature, 265, 726–728.
M., Bailey, A., & Rutter, M. (1994). A case-control family history Folstein, S., & Rutter, M. (1977b). Infantile autism: A genetic study of
study of autism. Journal of Child Psychology and Psychiatry, 35, 21 pairs. Journal of Child Psychology and Psychiatry, 18, 297–321.
877–900. Fombonne, E., Bolton, P., Prior, J., Jordan, H., & Rutter, M. (1997). A
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

Genetic Studies of Autism 13

family study of autism: Cognitive patterns and levels in parents and 1339–1345.
siblings. Journal of Child Psychology and Psychiatry, 38, 667–683. Lander, E. S. (1999). Array of hope. Nature Genetics, 21, 3–4.
Frith, U., & Happé, F. (1994). Autism: Beyond “theory of mind.” Cog- Le Couteur, A., Bailey, A. J., Goode, S., Pickles, A., Robertson, S.,
nition, 50, 115–132. Gottesman, I., & Rutter, M. (1996). A broader phenotype of autism:
Fulker, D. W., & Cherny, S. S. (1996). An improved multipoint sib-pair The clinical spectrum in twins. Journal of Child Psychology and
analysis of quantitative traits. Behavior Genetics, 26, 527–532. Psychiatry, 37, 785–801.
Gayán, J., Smith, S. D., Cherny, S. S., Cardon, L. R., Fulker, D. W., Le Couteur, A., Rutter, M., Lord, C., Rios, P., Robertson, S., Hold-
Brower, A. M., Olson, R. K., Pennington, B. F., & DeFries, J. C. grafer, M., & McLennan, J. (1989). Autism Diagnostic Interview:
(1999). Quantitative trait locus for specific language and reading A standardized investigator-based instrument. Journal of Autism
deficits on chromosome 6p. American Journal of Human Genetics, and Developmental Disorders, 19, 363–387.
64, 157–164. Lord, C., Rutter, M., Goode, S., Heemsbergen, J., Jordan, H., & Maw-
Gillberg, C., & Wahlström, J. (1985). Chromosome abnormalities in hood, L. (1989). Autism Diagnostic Observation Schedule: A stan-
infantile autism and other childhood psychoses: A population study dardized observation of communicative and social behavior. Jour-
of 66 cases. Developmental Medicine and Child Neurology, 27, nal of Autism and Developmental Disorders, 24, 659–685.
293–304. Maestrini, E., Lai, C., Marlow, A., Matthews, N., Wallace, S., Bailey, A.,
Goodman, R. (1991). Growing together and growing apart: The non- Cook, E.,Weeks, D. E., Monaco, A. P., & International Molecular
genetic forces on children in the same family. In P. McGuffin and Genetic Study of Autism Consortium (1999). Serotonin transporter
R. Murray (Eds.), The new genetics of mental illness (pp. 212–224). (5-HTT) and g-aminobutyric acid receptor subunit b3 (GABRB3)
Oxford, England: Heinemann Medical. gene polymorphisms are not associated with autism in the IMGSA
Griffith, E. M., Pennington, B. F., Wehner, E. A., & Rogers, S. J. (1999). families. American Journal of Medical Genetics (Neuropsychiatric
Executive functions in young children with autism. Child Develop- Genetics), 88, 492–496.
ment, 70, 817–832. Maestrini, E., Marlow, A. J., Weeks, D. E., & Monaco, A. P. (1998).
Grigorenko, E. L., Wood, F. B., Meyer, M. S., Hart, L. A., Speed, W. C., Molecular genetic investigations of autism. Journal of Autism and
Shuster, A., & Pauls, D. L. (1997). Susceptibility loci for distinct Developmental Disorders, 28, 439–449.
components of developmental dyslexia on chromosomes 6 and 15. Malhotra, A. K., & Goldman, D. (1999). Benefits and pitfalls encoun-
American Journal of Human Genetics, 60, 27–39. tered in psychiatric genetic association studies. Biological Psychi-
Gurling, H. M. D., Bolton, P. F., Vincent, J., Melmer, G., & Rutter, M. atry, 45, 544–550.
(1997). Molecular and cytogenetic investigations of the fragile X Merriman, T. R., Eaves, I. A., Twells, R. C. J., Merriman, M. E., Danoy,
region including the Frax A and Frax E CGG trinucleotide repeat P. A. C., Muxworthy, C. E., Hunter, K. M. D., Cox, R. D., Cucca,
sequences in families multiplex for autism and related phenotypes. F., McKinney, P. A., Shield, J. P. H., Baum, J. D., Tuomilehto, J.,
Human Heredity, 47, 254–262. Tuomilehto-Wolf, E., Ionesco-Tirgoviste, C., Joner, G., Thorsby, E.,
Hanson, D. R., & Gottesman, I. (1976). The genetics, if any, of infan- Undlien, D. E., Pociot, F., Nerup, J., Rønningen, K. S., Bain, S. C.,
tile autism and childhood schizophrenia. Journal of Autism and & Todd, J. A. (1997). Transmission of haplotypes of microsatel-
Schizophrenia, 6, 209–233. lite markers rather than single marker alleles in the mapping of
Happé, F. G. E. (1994). An advanced test of theory of mind: Under- a putative type 1 diabetes susceptibility gene (IDDM6). Human
standing of story characters’ thoughts and feelings by able autistic, Molecular Genetics, 7, 517–524.
mentally handicapped and normal children and adults. Journal of Merriman, T., Twells, R., Merriman, M., Eaves, I., Cox, R., Cucca, F.,
Autism and Developmental Disorders, 24, 129–154. McKinney, P., Shield, J., Baum, D., Bosi, E., Pozzilli, P., Nisticò,
Heavey, L., Phillips, W., Baron-Cohen, S., & Rutter, M. (in press). The L., Buzzetti, R., Joner, G., Rønningen, K., Thorsby, E., Undlien, D.,
Awkward Moments Test: A naturalistic measure of social under- Pociot, F., Nerup, J., Bain, S., Barnett, A., & Todd, J. (1998). Evi-
standing in autism. Journal of Autism and Developmental Disor- dence by allelic association-dependent methods for a type 1 diabetes
ders. polygene (IDDM6) on chromosome 18q21. Human Molecular Ge-
Hughes, C., Leboyer, M., & Bouvard, M. (1997). Executive function in netics, 6, 1003–1010.
parents of children with autism. Psychological Medicine, 27, 209– Meyers, C., Elias, S., & Arrabal, P. (1995). Congenital anomalies and
220. pregnancy loss. In L. G. Keith, E. Papiernik, D. M. Keith, & B. Luke
Hughes, C., Plumet, M.-H., & Leboyer, M. (1999). Towards a cognitive (Eds.), Multiple pregnancy: Epidemiology, gestation and perinatal
phenotype for autism: Increased prevalence of executive dysfunc- outcome (pp. 73–92). New York: Parthenon.
tion and superior spatial span amongst siblings of children with Molenaar, P. C. M., Boomsma, D. I., & Dolan, C. V. (1993). A third
autism. Journal of Child Psychology and Psychiatry, 40, 705–718. source of developmental differences. Behavior Genetics, 23, 519–
International Molecular Genetic Study of Autism Consortium (1998). A 524.
full genome screen for autism with evidence for linkage to a region Murphy, M., Bolton, P., Pickles, A. Fombonne, E., Piven, J., & Rutter,
on chromosome 7q. Human Molecular Genetics, 7, 571–578. M. (1999). Personality traits in the relatives of autistic probands.
Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Manuscript submitted for publication.
Child, 2, 217–250. Ozonoff, S., Rogers, S. J., Farnham, J. M., & Pennington, B. F. (1993).
Klauck, S. M., Poustka, F., Benner, A., Lesch, K.-P., & Poustka, A. Can standard measures identify subclinical markers of autism?
(1997). Serotonin transporter (5-HTT) gene variants associated with Journal of Autism and Developmental Disorders, 23, 429–441.
autism? Human Molecular Genetics, 6, 2244–2248. Philippe, A., Martinez, M., Guilloud-Bataille, M., Gillberg, C., Råstam,
Kruglyak, L., & Lander, E. S. (1995). Complete multipoint sib-pair anal- M., Sponterm, E., Coleman, M., Zapella, M., Aschauser, H., van
ysis of qualitative and quantitative traits. American Journal of Hu- Malldergerme, L., Penet, C., Feingoll, J., Brice, A., Leboyer,
man Genetics, 57, 439–454. M., & Paris Autism Research International Sibpair Study (1999).
Kurnit, D. M., Layton, W. M., & Matthysse, S. (1987). Genetics, chance Genome-wide scan for autism susceptibility genes. Human Molec-
and morphogenesis. American Journal of Human Genetics, 41, ular Genetics, 8, 805–812.
979–995. Pickles, A., Bolton, P., Macdonald, H., Bailey, A., Le Couteur, A., Sim,
Landa, R., Piven, J., Wzorek, M., Gayle, J. O., Chase, G. A., & Folstein, L., & Rutter, M. (1995). Latent class analysis of recurrence risk for
S. E. (1992). Social language use in parents of autistic individuals. complex phenotypes with selection and measurement error: A twin
Psychological Medicine, 22, 245–254. and family history study of autism. American Journal of Human
Landa, R., Wzorek, M., Piven, J., Folstein, S. E., & Isaacs, C. (1991). Genetics, 57, 717–726.
Spontaneous narrative discourse characteristics of parents of autis- Pickles, A., Starr, E., Kazak, S., Bolton, P., Papanikolau, K., Bailey,
tic individuals. Journal of Speech and Hearing Research, 34, A. J., Goodman, R., & Rutter, M. (in press). Variable expression of
P1: FDR
Journal of Abnormal Child Psychology PL114-45 February 2, 2000 15:53 Style file version June 09, 1999

14 Rutter

the autism broader phenotype: Findings from extended pedigrees. known medical conditions: Myth and substance. Journal of Child
Journal of Child Psychology and Psychiatry. Psychology and Psychiatry, 35, 311–322.
Piven, J., Chase, G. A., Landa, R., Wzorek, M., Gayle, J., Cloud, D., & Rutter, M., Dunn, J., Plomin, R., Simonoff, E., Pickles, A., Maughan,
Folstein, S. (1991). Psychiatric disorders in the parents of autistic B., Ormel, J., Meyer, J., & Eaves, L. (1997). Integrating nature
individuals. Journal of the American Academy of Child and Ado- and nurture: Implications of person-environment correlations and
lescent Psychiatry, 30, 471–478. interactions of developmental psychopathology. Development and
Piven, J., Gayle, J., Chase, J., Fink, B., Landa, R., Wrozek, M. & Folstein, Psychopathology, 9, 335–364.
S. (1990). A family history study of neuropsychiatric disorders in Rutter, M., Silberg, J., O’Connor, T., & Simonoff, E. (1999a). Genetics
the adult siblings of autistic individuals. Journal of the American and child psychiatry: I. Advances in quantitative and molecular
Academy of Child and Adolescent Psychiatry, 29, 177–183. genetics. Journal of Child Psychology and Psychiatry, 40, 3–18.
Piven, J., & Palmer, P. (1997). Cognitive deficits in parents from multiple- Rutter, M., Silberg, J., O’Connor, T., & Simonoff, E. (1999b). Genetics
incidence autism families. Journal of Child Psychology and Psy- and child psychiatry: II. Empirical research findings. Journal of
chiatry, 38, 1011–1022. Child Psychology and Psychiatry, 40, 19–55.
Piven, J., Palmer, P., Jacobi, D., Childress, D., & Arndt, S. (1997). Schroer, R. J., Phelan, M. C., Michaelis, R. C., Crawford, E. C., Skinner,
The broader autism phenotype: Evidence from a family study of S. A., Cuccaro, M., Simensen, R. J., Bishop, J., Skinner, C., Fender,
multiple-incidence autism families. American Journal of Psychia- D., & Stevenson, R. E. (1998). Autism and maternally derived aber-
try, 154, 185–190. rations of chromosome 15q. American Journal of Medical Genetics,
Piven, J., Wzorek, M., Landa, R., Lainhart, J., Bolton, P., Chase, G. A., 76, 327–336.
& Folstein, S. (1994). Personality characteristics of the parents of Sibilia, M., & Wagner, E. F. (1996). Transgenic animals. European Re-
autistic individuals. Psychological Medicine, 24, 783–795. view, 4, 371–392.
Pocock, S. J. (1983). Clinical trials: A practical approach. Chichester, Smalley, S. L. (1998). Autism and tuberous sclerosis. Journal of Autism
England: Wiley. and Developmental Disorders, 28, 407–414.
Risch, N., & Merikangas, K. (1996). The future of genetic studies of Smalley, S. L., McCracken, J., & Tanguay, P. (1995). Autism, affective
complex human diseases. Science, 273, 1516–1517. disorders and social phobia. American Journal of Medical Genetics,
Rutter, M. (1967). Psychotic disorders in early childhood. In A. J. Coppen 60, 19–26.
and A. Walk (Eds.), Recent Developments in Schizophrenia (pp. Spielman, R. S., & Ewens, W. J. (1996). Invited editorial: The TDT and
133–158). Ashford, Kent: Headley Bros/RMPA. other family-based tests for linkage disequilibrium and association.
Rutter, M. (1994). Psychiatric genetics: Research challenges and path- American Journal of Human Genetics, 59, 983–989.
ways forward. American Journal of Medical Genetics (Neuropsy- Starr, L., Kazak, S., Pickles, A., Tomlins, M., Bailey, A., & Rutter, M.
chiatric Genetics), 54, 185–198. (1999). Family genetic study of autism associated with profound
Rutter, M. (1999a). Psychosocial adversity and child psychopathology. mental retardation. Submitted for publication.
British Journal of Psychiatry, 174, 480–493. Steffenburg, S., Gillberg, C., Hellgren, L., Andersson, L., Gillberg, I.,
Rutter, M. (1999b). The Emanuel Miller Memorial Lecture 1998. Jakobsson, G., & Bohman, M. (1989). A twin study of autism in
Autism: Two-way interplay between research and clinical work. Denmark, Finland, Iceland, Norway and Sweden. Journal of Child
Journal of Child and Adolescent Psychiatry, 40, 169–188. Psychology and Psychiatry, 30, 405–416.
Rutter, M., Andersen-Wood, L., Beckett, C., Bredenkamp, D., Wahlsten, D. (1999). Single-gene influences on brain and behavior. An-
Castle, J., Groothues, C., Kreppner, J., Keaveney, L., Lord, C., nual Review of Psychology, 50, 599–624.
O’Connor, T. G., & English and Romanian Adoptee (ERA) Watson, S. J., & Akil, H. (1999). Gene chips and arrays revealed: A
Study Team (1999). Quasi-autistic patterns following severe early primer on their power and their uses. Biological Psychiatry, 45,
global privation. Journal of Child Psychology and Psychiatry, 40, 533–543.
537–549. Watson, J. D., Gilman, M., Witkowski, J., & Zoller, M. (1992). Recom-
Rutter, M., Bailey, A., Bolton, P., & Le Couteur, A. (1994). Autism and binant DNA (2nd ed.). New York: Scientific American Books.