Part III Origins, Patterns, and Prognoses Dyslexia researchers have long been deeply interested in causation with some of the most challenging work in this realm occurring in genetic and familial studies. Sandra Smith looked at the extended families of 12 young adults to determine the degree to which learning disabilities and/or related disorders occurred in families. "Familial Patterns of Learning Disabilities" reports that there is a strong familial incidence of learning disabilities, although the specific disability appears not to be inherited. Smith also found a significant correlation between learning disabilities and autoimmune disease. Nathlie A. Badian continues her investigations into the relationship between right hemisphere dysfunction and behavior. "Nonverbal Learning Disability, School Behavior, and Dyslexia" details her most recent findings. Badian concludes that students displaying low nonverbal, high verbal profiles who appear to be dyslexic are at greater risk for social problems and school failure than poor readers whose profile shows high nonverbal, low verbal abilities. This section also contains reports of several research studies which analyze elements of language and the role that these elements play in mastering reading and writing. Much of the cutting-edge research being done in ths area derives, if not directly at least in part, from the work of Isabel Liberman and her colleagues at Haskins Laboratory, as well as from several British and European researchers. A number of studies have noted that children who have difficulty with naming objects are likely to have reading and spelling problems as well. Alisa Cantwell and Hyla Rubin report on the effect of object naming on the written language of adults. In "Object Naming Ability of Adults with Written Language Difficulties," they report that adults who are poor readers and spellers exhibit phonological deficits and lack understanding of orthography. Margaret L. Stanback analyzed over 17,000 frequently-used words, grouping them by syllable patterns for teaching reading. "Syl141
142
ORIGINS, PAI"rERNS, AND PROGNOSES
lable and Rime Patterns for Teaching Reading: Analysis of a FrequencyBased Vocabulary of 17,602 Words" presents a strong case for the usefulness of onsets and rimes in teaching reading. Stanback notes that associating written letters with sounds is the foundation for understanding the alphabetic principle. Her study further indicates that taking advantage of common rime patterns can be a powerful tool for enhancing students' reading vocabularies. According to Liana K. Holt-Ochsner, fluency in reading and improved comprehension depend on word recognition skills that have been mastered to the point of automaticity. Her paper, "Automaticity Training for Dyslexics: An Experimental Stud~" describes the testing of the hypothesis that increasing word recognition so that it is rapid and automatic results in improved reading comprehension. That phonological awareness is critical to mastering the alphabetic principle and, hence, to developing skill in reading has been attested to by a number of major research studies over the past decade. Patricia C. Lindamood and her associates look at some of the subtle factors affecting ability to self-correct and to master decoding and spelling skills. "Issues in Assessment of Phonological Awareness" focuses particularly on the comparator function, comparing how and where two words differ. The authors call for refining the means of measuring this function as critical to diagnosis and to selecting appropriate interventions in reading and spelling.
Familial Patterns of Learning Disabilities Sandra Smith
SequimSchoolDistrict Sequim,Washington
Extended families of 12 young adults (9 LD, 3 non-LD) were given a battery of tests and questionnaires, and 131 persons, ranging in age from 6 to 85, were classified as LD or non-LD on the basis of subtest scores I SD below the mean or less on subtests of the PlAT and WRAT achievement tests. Pedigree analysis indicated that LD was strongly familial, with the most probable mode involving a major gene effect, but the type of disability (reading/math) was not directly inherited. Autoimmune disorders were significantly correlated (P < .005) with LD, especially in families in which LD remained a major handicap into adulthood, a trait that also varied between families. In two of the LD families, adults showed little evidence of the reading~spelling deficits they had shown when tested as children, while adults in other families failed to make gains in reading and spelling.
Large extended families provide valuable information about the inheritance patterns of complex traits, especially those conditions for which previous studies indicate that the population includes those with different etiologies (Emery 1986; Lubs et al. 1991). When persons from several generations are available, they can be examined for changes in the manifestation of the trait over time, and for the development of diseases that may accompany the trait, but which do not become apparent until middle age. Learning disabilities represent this type of trait, in which different causes of the trait are assumed, but for which the genetic mechanisms of the various causes have not been differentiated (Lewitter, DeFries, and Elston 1980; Smith and Goldgar Annalsd Dyslexia,Vol.42, 1992. Copyright© 1992by The Orton DyslexiaSociety ISSN 0736-9387 143
144
OmcxNs, P,4vre,P,NS, AND PROGNOSES
1986). Nuclear family studies such as Borges-Osorio and Salzano (1987), Owen (1978), and the extensive Colorado Family Reading Study (DeFries and Decker 1982) indicate that the parents and siblings of children diagnosed with reading disabilities were also more likely to be affected by poor reading and spelling skills. The genetic mechanism for this pattern was unclear, however, with earlier studies by Hallgren (1950), Op't Hof and Guldenpfennig (1972), and Zahalkova, Vrzal, and Kloboukova (1972) proposing dominant inheritance, while Sladen (1972) hypothesized that females needed two carrier parents to be affected themselves, but males would be reading disabled with only one such gene. More recently, there have been several investigations of the inheritance of component processing skills within learning-disabled families, as well as of severity and subtype characteristics. Finucci and Childs (1983) developed a test battery suitable for both children and adults with which they compared the severity of dyslexia within nuclear families. While the number of affected children increased when both parents were clearly affected, the severity level of reading disability did not show a high correlation between the two generations. After Lewitter, DeFries, and Elston (1980) were unable to determine a single genetic model that could account for the observed inheritance patterns in their subgroups of 125 nuclear families of reading-disabled children, the Colorado Family Reading Study looked at subtypes based on reading, spatial/reasoning, and symbol-processing speed components. While they found strong indications of familial transmission of reading and spelling difficulties, they saw no evidence for sex linkage or direct inheritance of the subtypes themselves. Even in monozygotic twins, Ho, Gilger, and Decker (1988) noted that only six out of nine were concordant for Bannatyne's "genetic dyslexia" subtype, in which spatial skills were affected. Omenn and Weber (1978) had previously examined the parent-sibling-proband correlation of reading disability subtypes, using a phonetic/whole word recognition dichotomy, concluding that dominant inheritance best explained their observations, but finding, as in the Colorado study, that parents were less likely than siblings to share probands' disability subtypes. The same lack of heritability of specific subtypes, in this case phonological versus orthographic, was demonstrated by Szeszulski and Manis in their 1990 report on 40 dyslexic children and their parents. They did, however, find that at least one parent in each family showed deficits in phonological processing, regardless of the subtype to which the affected child was assigned, and 82 percent of the children with such difficulties in spelling-to-sound correspondences also had at least one parent so affected. Gordon (1980) and Regehr and Kaplan (1988) reported familial similarities for
FAMILIAL PATTERNS OF LEARNING DISABILITIES
145
some reading disabilities, but, again, could not substantiate a direct inheritance of subtypes. Studies of extended families have been less common, but Elbert and Seale's 1988 pedigree of 16 members from one family supported autosomal dominant inheritance of a gene which they suggested might have "variable effects on neuroanatomical structure during brain development," and brain imaging studies on two of the family members did indeed show non-identical abnormalities near language centers. Lubs et al. (1991) are also using MRI and PET scans to study the localization of brain differences in some of their ten extended families in which dyslexia has occurred over at least three generations. While their research is still in process, preliminary data indicate that, in these persons whose relatively pure spelling and reading problems appear to be dominantly inherited, some differences exist in the structures of several parts of the brain involved with language and visual processing. Specific brain abnormalities form the crux of what is known as the "Geschwind hypothesis," the theory first proposed by Geschwind and Behan (1982) that attributed an association between left-handedness, autoimmune disorders, and language disabilities, to the effects of testosterone on the developing fetal brain. While more recent studies have failed to find significant correlations between dyslexia in students (and in their nuclear families) and an overall increased incidence of immune system disorders, asthma has been noted to occur more frequently in persons with verbal deficits or dyslexia (DeFries et al. 1991; Pennington et al. 1987; Smith, Meyers, and Kline 1989). Among more serious autoimmune dysfunctions, Lahita (1988) noted significantly more LD in male relatives of his patients with systemic lupus erythematosus (SLE) than in relatives of controls. In fact, of 109 female SLE patients, 55 of their 90 children had LD, primarily dyslexia. In the hypothesis of Arnett, Bias, and Reveille (1989), a single gene would convey susceptibility to autoimmunity, with organ specificity determined by other genes; therefore concentrations of autoimmunity and related disorders would be more prevalent in certain ethnic groups which share such genes in common. The mechanism by which a single gene could cause autoimmunity is still in doubt, however. Several studies (i.e. Shoenfeld et al. 1989; Theofilopoulos and Kofler 1989) implicate the variable regions of T cell antigen receptors in colitis, Crohn's disease, and SLE, but an exact correspondence between genetic variability and disease is not yet established. Interestingly, however, while Lubs et al. (1991) found no increase in autoimmunity within the dyslexic members of his extended families, he did note that the majority of females carrying the gene for dyslexia had a increased ratio of helper T cells to suppressor T cells; the same relative deficiency of suppressor
146
ORzGZNS, PAVrE~S, AND PROGNOSBS
T cells was also reported by Warren et al. (1990) to occur in autistic females. Whether this ratio difference is somehow connected to greater immunological reactivit3¢ such as seen in allergy and autoimmunity, is unclear. With many researchers suggesting that LD is a heterogeneous disorder, studies of extended families can clarify the inheritance of LD because all affected persons within a family are likely to share the same key genes. For example, in 1983, Smith et al. identified a subset of Midwest American LD families as having a gene for dyslexia linked to chromosome 15, yet Bisgaard et al. (1987) found no such linkage among their Danish extended families, indicating heterogeneity between ethnic groups. Since then, Smith and her colleagues (DeFries et al. 1991) have expanded their linkage testing, finding that only a small subset of their families appears to have dyslexia linked to chromosome 15, while another subset may link to chromosome 6. Taken together, however, these previous studies do support the hypothesis that dyslexia is a symptom of a genetically determined neurological disorder with immunological components (Galaburda, Rosen, and Sherman 1989). The present study examines nine extended families in which one person had been diagnosed when a child simply as having "learning disabilities" in a broad sense, including dyscalcula and combinations of math, spelling, and reading disorders. A test battery and questionnaires classified relatives as LD or non-LD, and identified biological correlates such as ear infections and autoimmune diseases that accompany LD within these families. The Method Subjects Adults who had been diagnosed as "Learning Disabled" under Washington State law between 1975 and 1988 in a small rural school district were asked to participate if they had at least ten relatives nearby who would be willing to be tested for "a study of learning patterns." The nine probands whose families agreed all had IQs in the normal range and included those with math disability, reading and/or spelling disability, and combinations of the three. Three control families with similar characteristics were enlisted from non-LD classmates of the probands, but one family proved to have LD extending through three generations, so its members were moved to the "LD Families" category for academic test reporting and eliminated from the segregation analysis. All families were of European ancestry, with Scandinavian, British, and German being most common, and most wage earners were skilled craftspersons. Families in which there was a known history of drug or
FAMILIAL PATTERNS OF LEARNINGDISABILrrIES
147
alcohol abuse were not asked to join the stud3~ but the only significant differences between participants and non-participants were size of family in the area and knowledge of other family members with learning problems, although both participating probands and nonparticipants included those who reported no other affected family members. Procedures
The basic test protocol was that used by Finucci et al. (1982) and Pennington et al. (1987). Unrevised forms of the PIAT (subtests of Reading Recognition, Reading Comprehension, Spelling, Mathematics and General Information), Wide Range Achievement Test (WRAT) Spelling, and Gray Oral Reading Test (GORT) were used. To determine if the subjects had adequate cognitive abilities for normal academic performance, they also took the Raven Progressive Matrices, a test of nonverbal reasoning. Three sets of questionnaires assessed handedness, health and development, and reading history. The Oldfield Handedness Inventory yields a Laterality Quotient of from - 100 (most left-handed) to + 100, based on self-reports of the hand most preferred for common tasks. Persons tested or their parents also completed several pages of questionnaire which asked about their previous school history, reading habits, and medically-diagnosed health conditions. Persons were diagnosed as LD or non-LD based on the PIAT and WRAT subtests for which preschool-to-adult norms and standard scores were available. "LD" persons were those with at least one subtest score of reading, math, or spelling falling one standard deviation (SD) or more below the mean at which the subtest was normed and with at least one other indication of a learning disability. The second indication could include other subtest scores below one standard deviation, reading history quotients above .40 (where .00 indicates no reading or school history problems), or oral reading/spelling subtest scores 1.5 standard deviations below math subtest scores. To examine whether dyslexia was inherited as a subtype of LD, a Reading Quotient (RQ) was determined by dividing the average of the Gray Oral Reading and WRAT Spelling age means by the average of chronological age, age for grade placement, and IQ age equivalence for children, and dividing by the average of age for last grade completed and age equivalents for the PIAT Math and General Information subtests for adults. Those persons with RQ scores above .80 were considered nondyslexic (Finucci 1978). All persons identified by this criterion as dyslexic were also classifted as "LD" on the basis of subtest standard scores falling less than 1.5 standard deviations below the published means. Out of 131 persons classified by testing, 64 were unaffected, 50 were dyslexic or borderline dyslexic, and six had learning disorders primarily affecting math skills.
148
O~aizvs, P A ~ s ~ r s , alvo PROC;ZvOSES
This left eleven "questionables" who did not meet all the criteria for LD, but who nonetheless reported past difficulties with written language. Since seven of these eleven obtained Raven Progressive Matrices (SPM) z-scores above 1.10, and four obtained the maximum score possible, they may be well-compensated adults. In fact, one at age 19 showed no sign of the spelling deficit which plagued him during previous testing at age ten. Except for one who was eliminated due to incomplete testing results, these "questionables" were considered as affected in the following analyses. An additional 31 persons for whom previous testing or sufficient other documentation was available were also classified on the pedigrees as LD or non-LD. The pedigrees of the nine original LD families (see Figure 1 for sample pedigrees) were then examined for evidence of assortive mating, Mendelian inheritance patterns, interfamilial differences, and the inheritance of specific disability subtypes, using both traditional pedigree analysis and complex segregation analysis.
Results Test Scores
Z-score means of the overall group (n = 131)closely approximated published means for most subtests. However, the PlAT Spelling subtest mean was - 0.571, probably reflecting the high percentage (43 percent) of persons with LD. The Raven SPM mean was 0.639, with much of the elevated scores contributed by those over 50. Given the active lifestyle of the senior citizens tested, the original norms, based on British senior-center visitors, may be invalid for this population. The 56 LD persons showed significantly lower subtest score means than the 64 non-LD persons tested for most subtests except on General Information and Raven's SPM. Although Gillis and DeFries (1989) had found the Reading Recognition subtest to be the best discriminator of reading disability in their revised test battery, PlAT Spelling proved more effident a predictor for this older group, possibly because longer reading practice has given them more extensive sight vocabularies. To determine if non-LD relatives of LD family members shared more characteristics with their affected relatives, or with unaffected non-relatives, subjects were further divided into 3 groups: LD family members (Group I), unaffected members of LD families (Group II), and non-LD persons not from LD families (Group III). Included in the 61 LD family members were ten of the "questionables" who had had language or reading problems at earlier ages. For all subtests, Group III had the highest mean scores. However, Group II scores resembled those of Group III much more closely than those of their affected rela-
F~u~I.
THYROID
01
.
i
01.0102
02.02.04
°Ti.,.i'h
tHYROID
ix ,t> / BRAIN 5
ii,,,,
/
A
01.03.12
01.
•
02.~
04.02.01
I
0t0219
Umflecl~l
ASTHMA
/,'-, ~/~. . . . . ASTHMA
[]
01.03.1'"-' 8"" '01.03,20
01.~:17
01.02,
6
DI/~EI~S
04.01.02 ~:~
.
.
i
04.03.~
[]
I
Learning Oueslionable-- Unclassifiable-Disabled (ProbablyLO) (Tooyoung,
•
-
x
Tested Previously lested
-
°'°~"i--~°'o~"
040303T040353 04.0354~0403. • 04. UN~=FECTm ] BY HIS~RY I
04.02.31
FAMILY4
~,o% ,l, ~, ,lo,o~,~
101.03.15 01.02.16
~ I
•
oo,_,~,sl"
02 02.[~
01.03.11
01.03.10
01,03,09
.~.02
101.03.08
01.03.071
01.0006
01.03.06
01.03.04
01
a, 6 6 6 ~:,o .~
0101.03
3
"1,~'~
0202.02
02.01.01
• ASTHMA
~01.02
Pedigrees of Three Families Circles = female; black symbols = learning disabled; lightly dotted symbols = not affected; heavily stipled symbols = tested, but not classified due to age below 6 or physical handicap. Symbols with * or x were given test battery including math, reading, and spelling assessments.
JRA
03
01.02.01
FmlLY1
2
~02.01
FAMILY2
150
ORrGxNS, PA~BaNS, AND PROClVOSES
tives in Group I. Using analysis of variance to compare the three groups across the academic subtest scores, z-score means for Groups II and III did not differ significantly (Scheffe's contrast method) for any academic subtest. On the other hand, the LD Group I means differed significantly from all subtest means of both Groups II and III, with the exception of PLAT General Information, in which Group I differed significantly (Scheffe F = 3.84) only between Groups I and III. PLATMath, Reading Recognition, Comprehension, and Spelling, as well as the WRAT Spelling subtest, all differed at the p
142
ORIGINS, PAI"rERNS, AND PROGNOSES
lable and Rime Patterns for Teaching Reading: Analysis of a FrequencyBased Vocabulary of 17,602 Words" presents a strong case for the usefulness of onsets and rimes in teaching reading. Stanback notes that associating written letters with sounds is the foundation for understanding the alphabetic principle. Her study further indicates that taking advantage of common rime patterns can be a powerful tool for enhancing students' reading vocabularies. According to Liana K. Holt-Ochsner, fluency in reading and improved comprehension depend on word recognition skills that have been mastered to the point of automaticity. Her paper, "Automaticity Training for Dyslexics: An Experimental Stud~" describes the testing of the hypothesis that increasing word recognition so that it is rapid and automatic results in improved reading comprehension. That phonological awareness is critical to mastering the alphabetic principle and, hence, to developing skill in reading has been attested to by a number of major research studies over the past decade. Patricia C. Lindamood and her associates look at some of the subtle factors affecting ability to self-correct and to master decoding and spelling skills. "Issues in Assessment of Phonological Awareness" focuses particularly on the comparator function, comparing how and where two words differ. The authors call for refining the means of measuring this function as critical to diagnosis and to selecting appropriate interventions in reading and spelling.
Familial Patterns of Learning Disabilities Sandra Smith
SequimSchoolDistrict Sequim,Washington
Extended families of 12 young adults (9 LD, 3 non-LD) were given a battery of tests and questionnaires, and 131 persons, ranging in age from 6 to 85, were classified as LD or non-LD on the basis of subtest scores I SD below the mean or less on subtests of the PlAT and WRAT achievement tests. Pedigree analysis indicated that LD was strongly familial, with the most probable mode involving a major gene effect, but the type of disability (reading/math) was not directly inherited. Autoimmune disorders were significantly correlated (P < .005) with LD, especially in families in which LD remained a major handicap into adulthood, a trait that also varied between families. In two of the LD families, adults showed little evidence of the reading~spelling deficits they had shown when tested as children, while adults in other families failed to make gains in reading and spelling.
Large extended families provide valuable information about the inheritance patterns of complex traits, especially those conditions for which previous studies indicate that the population includes those with different etiologies (Emery 1986; Lubs et al. 1991). When persons from several generations are available, they can be examined for changes in the manifestation of the trait over time, and for the development of diseases that may accompany the trait, but which do not become apparent until middle age. Learning disabilities represent this type of trait, in which different causes of the trait are assumed, but for which the genetic mechanisms of the various causes have not been differentiated (Lewitter, DeFries, and Elston 1980; Smith and Goldgar Annalsd Dyslexia,Vol.42, 1992. Copyright© 1992by The Orton DyslexiaSociety ISSN 0736-9387 143
144
OmcxNs, P,4vre,P,NS, AND PROGNOSES
1986). Nuclear family studies such as Borges-Osorio and Salzano (1987), Owen (1978), and the extensive Colorado Family Reading Study (DeFries and Decker 1982) indicate that the parents and siblings of children diagnosed with reading disabilities were also more likely to be affected by poor reading and spelling skills. The genetic mechanism for this pattern was unclear, however, with earlier studies by Hallgren (1950), Op't Hof and Guldenpfennig (1972), and Zahalkova, Vrzal, and Kloboukova (1972) proposing dominant inheritance, while Sladen (1972) hypothesized that females needed two carrier parents to be affected themselves, but males would be reading disabled with only one such gene. More recently, there have been several investigations of the inheritance of component processing skills within learning-disabled families, as well as of severity and subtype characteristics. Finucci and Childs (1983) developed a test battery suitable for both children and adults with which they compared the severity of dyslexia within nuclear families. While the number of affected children increased when both parents were clearly affected, the severity level of reading disability did not show a high correlation between the two generations. After Lewitter, DeFries, and Elston (1980) were unable to determine a single genetic model that could account for the observed inheritance patterns in their subgroups of 125 nuclear families of reading-disabled children, the Colorado Family Reading Study looked at subtypes based on reading, spatial/reasoning, and symbol-processing speed components. While they found strong indications of familial transmission of reading and spelling difficulties, they saw no evidence for sex linkage or direct inheritance of the subtypes themselves. Even in monozygotic twins, Ho, Gilger, and Decker (1988) noted that only six out of nine were concordant for Bannatyne's "genetic dyslexia" subtype, in which spatial skills were affected. Omenn and Weber (1978) had previously examined the parent-sibling-proband correlation of reading disability subtypes, using a phonetic/whole word recognition dichotomy, concluding that dominant inheritance best explained their observations, but finding, as in the Colorado study, that parents were less likely than siblings to share probands' disability subtypes. The same lack of heritability of specific subtypes, in this case phonological versus orthographic, was demonstrated by Szeszulski and Manis in their 1990 report on 40 dyslexic children and their parents. They did, however, find that at least one parent in each family showed deficits in phonological processing, regardless of the subtype to which the affected child was assigned, and 82 percent of the children with such difficulties in spelling-to-sound correspondences also had at least one parent so affected. Gordon (1980) and Regehr and Kaplan (1988) reported familial similarities for
FAMILIAL PATTERNS OF LEARNING DISABILITIES
145
some reading disabilities, but, again, could not substantiate a direct inheritance of subtypes. Studies of extended families have been less common, but Elbert and Seale's 1988 pedigree of 16 members from one family supported autosomal dominant inheritance of a gene which they suggested might have "variable effects on neuroanatomical structure during brain development," and brain imaging studies on two of the family members did indeed show non-identical abnormalities near language centers. Lubs et al. (1991) are also using MRI and PET scans to study the localization of brain differences in some of their ten extended families in which dyslexia has occurred over at least three generations. While their research is still in process, preliminary data indicate that, in these persons whose relatively pure spelling and reading problems appear to be dominantly inherited, some differences exist in the structures of several parts of the brain involved with language and visual processing. Specific brain abnormalities form the crux of what is known as the "Geschwind hypothesis," the theory first proposed by Geschwind and Behan (1982) that attributed an association between left-handedness, autoimmune disorders, and language disabilities, to the effects of testosterone on the developing fetal brain. While more recent studies have failed to find significant correlations between dyslexia in students (and in their nuclear families) and an overall increased incidence of immune system disorders, asthma has been noted to occur more frequently in persons with verbal deficits or dyslexia (DeFries et al. 1991; Pennington et al. 1987; Smith, Meyers, and Kline 1989). Among more serious autoimmune dysfunctions, Lahita (1988) noted significantly more LD in male relatives of his patients with systemic lupus erythematosus (SLE) than in relatives of controls. In fact, of 109 female SLE patients, 55 of their 90 children had LD, primarily dyslexia. In the hypothesis of Arnett, Bias, and Reveille (1989), a single gene would convey susceptibility to autoimmunity, with organ specificity determined by other genes; therefore concentrations of autoimmunity and related disorders would be more prevalent in certain ethnic groups which share such genes in common. The mechanism by which a single gene could cause autoimmunity is still in doubt, however. Several studies (i.e. Shoenfeld et al. 1989; Theofilopoulos and Kofler 1989) implicate the variable regions of T cell antigen receptors in colitis, Crohn's disease, and SLE, but an exact correspondence between genetic variability and disease is not yet established. Interestingly, however, while Lubs et al. (1991) found no increase in autoimmunity within the dyslexic members of his extended families, he did note that the majority of females carrying the gene for dyslexia had a increased ratio of helper T cells to suppressor T cells; the same relative deficiency of suppressor
146
ORzGZNS, PAVrE~S, AND PROGNOSBS
T cells was also reported by Warren et al. (1990) to occur in autistic females. Whether this ratio difference is somehow connected to greater immunological reactivit3¢ such as seen in allergy and autoimmunity, is unclear. With many researchers suggesting that LD is a heterogeneous disorder, studies of extended families can clarify the inheritance of LD because all affected persons within a family are likely to share the same key genes. For example, in 1983, Smith et al. identified a subset of Midwest American LD families as having a gene for dyslexia linked to chromosome 15, yet Bisgaard et al. (1987) found no such linkage among their Danish extended families, indicating heterogeneity between ethnic groups. Since then, Smith and her colleagues (DeFries et al. 1991) have expanded their linkage testing, finding that only a small subset of their families appears to have dyslexia linked to chromosome 15, while another subset may link to chromosome 6. Taken together, however, these previous studies do support the hypothesis that dyslexia is a symptom of a genetically determined neurological disorder with immunological components (Galaburda, Rosen, and Sherman 1989). The present study examines nine extended families in which one person had been diagnosed when a child simply as having "learning disabilities" in a broad sense, including dyscalcula and combinations of math, spelling, and reading disorders. A test battery and questionnaires classified relatives as LD or non-LD, and identified biological correlates such as ear infections and autoimmune diseases that accompany LD within these families. The Method Subjects Adults who had been diagnosed as "Learning Disabled" under Washington State law between 1975 and 1988 in a small rural school district were asked to participate if they had at least ten relatives nearby who would be willing to be tested for "a study of learning patterns." The nine probands whose families agreed all had IQs in the normal range and included those with math disability, reading and/or spelling disability, and combinations of the three. Three control families with similar characteristics were enlisted from non-LD classmates of the probands, but one family proved to have LD extending through three generations, so its members were moved to the "LD Families" category for academic test reporting and eliminated from the segregation analysis. All families were of European ancestry, with Scandinavian, British, and German being most common, and most wage earners were skilled craftspersons. Families in which there was a known history of drug or
FAMILIAL PATTERNS OF LEARNINGDISABILrrIES
147
alcohol abuse were not asked to join the stud3~ but the only significant differences between participants and non-participants were size of family in the area and knowledge of other family members with learning problems, although both participating probands and nonparticipants included those who reported no other affected family members. Procedures
The basic test protocol was that used by Finucci et al. (1982) and Pennington et al. (1987). Unrevised forms of the PIAT (subtests of Reading Recognition, Reading Comprehension, Spelling, Mathematics and General Information), Wide Range Achievement Test (WRAT) Spelling, and Gray Oral Reading Test (GORT) were used. To determine if the subjects had adequate cognitive abilities for normal academic performance, they also took the Raven Progressive Matrices, a test of nonverbal reasoning. Three sets of questionnaires assessed handedness, health and development, and reading history. The Oldfield Handedness Inventory yields a Laterality Quotient of from - 100 (most left-handed) to + 100, based on self-reports of the hand most preferred for common tasks. Persons tested or their parents also completed several pages of questionnaire which asked about their previous school history, reading habits, and medically-diagnosed health conditions. Persons were diagnosed as LD or non-LD based on the PIAT and WRAT subtests for which preschool-to-adult norms and standard scores were available. "LD" persons were those with at least one subtest score of reading, math, or spelling falling one standard deviation (SD) or more below the mean at which the subtest was normed and with at least one other indication of a learning disability. The second indication could include other subtest scores below one standard deviation, reading history quotients above .40 (where .00 indicates no reading or school history problems), or oral reading/spelling subtest scores 1.5 standard deviations below math subtest scores. To examine whether dyslexia was inherited as a subtype of LD, a Reading Quotient (RQ) was determined by dividing the average of the Gray Oral Reading and WRAT Spelling age means by the average of chronological age, age for grade placement, and IQ age equivalence for children, and dividing by the average of age for last grade completed and age equivalents for the PIAT Math and General Information subtests for adults. Those persons with RQ scores above .80 were considered nondyslexic (Finucci 1978). All persons identified by this criterion as dyslexic were also classifted as "LD" on the basis of subtest standard scores falling less than 1.5 standard deviations below the published means. Out of 131 persons classified by testing, 64 were unaffected, 50 were dyslexic or borderline dyslexic, and six had learning disorders primarily affecting math skills.
148
O~aizvs, P A ~ s ~ r s , alvo PROC;ZvOSES
This left eleven "questionables" who did not meet all the criteria for LD, but who nonetheless reported past difficulties with written language. Since seven of these eleven obtained Raven Progressive Matrices (SPM) z-scores above 1.10, and four obtained the maximum score possible, they may be well-compensated adults. In fact, one at age 19 showed no sign of the spelling deficit which plagued him during previous testing at age ten. Except for one who was eliminated due to incomplete testing results, these "questionables" were considered as affected in the following analyses. An additional 31 persons for whom previous testing or sufficient other documentation was available were also classified on the pedigrees as LD or non-LD. The pedigrees of the nine original LD families (see Figure 1 for sample pedigrees) were then examined for evidence of assortive mating, Mendelian inheritance patterns, interfamilial differences, and the inheritance of specific disability subtypes, using both traditional pedigree analysis and complex segregation analysis.
Results Test Scores
Z-score means of the overall group (n = 131)closely approximated published means for most subtests. However, the PlAT Spelling subtest mean was - 0.571, probably reflecting the high percentage (43 percent) of persons with LD. The Raven SPM mean was 0.639, with much of the elevated scores contributed by those over 50. Given the active lifestyle of the senior citizens tested, the original norms, based on British senior-center visitors, may be invalid for this population. The 56 LD persons showed significantly lower subtest score means than the 64 non-LD persons tested for most subtests except on General Information and Raven's SPM. Although Gillis and DeFries (1989) had found the Reading Recognition subtest to be the best discriminator of reading disability in their revised test battery, PlAT Spelling proved more effident a predictor for this older group, possibly because longer reading practice has given them more extensive sight vocabularies. To determine if non-LD relatives of LD family members shared more characteristics with their affected relatives, or with unaffected non-relatives, subjects were further divided into 3 groups: LD family members (Group I), unaffected members of LD families (Group II), and non-LD persons not from LD families (Group III). Included in the 61 LD family members were ten of the "questionables" who had had language or reading problems at earlier ages. For all subtests, Group III had the highest mean scores. However, Group II scores resembled those of Group III much more closely than those of their affected rela-
F~u~I.
THYROID
01
.
i
01.0102
02.02.04
°Ti.,.i'h
tHYROID
ix ,t> / BRAIN 5
ii,,,,
/
A
01.03.12
01.
•
02.~
04.02.01
I
0t0219
Umflecl~l
ASTHMA
/,'-, ~/~. . . . . ASTHMA
[]
01.03.1'"-' 8"" '01.03,20
01.~:17
01.02,
6
DI/~EI~S
04.01.02 ~:~
.
.
i
04.03.~
[]
I
Learning Oueslionable-- Unclassifiable-Disabled (ProbablyLO) (Tooyoung,
•
-
x
Tested Previously lested
-
°'°~"i--~°'o~"
040303T040353 04.0354~0403. • 04. UN~=FECTm ] BY HIS~RY I
04.02.31
FAMILY4
~,o% ,l, ~, ,lo,o~,~
101.03.15 01.02.16
~ I
•
oo,_,~,sl"
02 02.[~
01.03.11
01.03.10
01,03,09
.~.02
101.03.08
01.03.071
01.0006
01.03.06
01.03.04
01
a, 6 6 6 ~:,o .~
0101.03
3
"1,~'~
0202.02
02.01.01
• ASTHMA
~01.02
Pedigrees of Three Families Circles = female; black symbols = learning disabled; lightly dotted symbols = not affected; heavily stipled symbols = tested, but not classified due to age below 6 or physical handicap. Symbols with * or x were given test battery including math, reading, and spelling assessments.
JRA
03
01.02.01
FmlLY1
2
~02.01
FAMILY2
150
ORrGxNS, PA~BaNS, AND PROClVOSES
tives in Group I. Using analysis of variance to compare the three groups across the academic subtest scores, z-score means for Groups II and III did not differ significantly (Scheffe's contrast method) for any academic subtest. On the other hand, the LD Group I means differed significantly from all subtest means of both Groups II and III, with the exception of PLAT General Information, in which Group I differed significantly (Scheffe F = 3.84) only between Groups I and III. PLATMath, Reading Recognition, Comprehension, and Spelling, as well as the WRAT Spelling subtest, all differed at the p
