Brain (1992), 115, 1807-1826

THE TOPOGRAPHY OF CALLOSAL READING PATHWAYS A CASE-CONTROL ANALYSIS by

j . R. BINDER and

J. P. MOHR

(From the Department of Neurology, The Neurological Institute, Columbia Presbyterian Medical Center, New York, USA)

Lesion topography and reading ability were analysed in 17 patients with dominant posterior cerebral artery territory infarction. Patients with dominant posterior cerebral artery infarction in whom reading was unaffected served as an anatomical control group. Normal readers had lesions in the medial and ventral occipital lobe, sparing dorsal white matter pathways and the ventral temporal lobe. Global and permanent alexia occurred only with additional injury to the splenium, forceps major or white matter above the occipital horn of the lateral ventricle. These data suggest that callosal pathways mediating reading lie above the occipital horn and have little connection with the ventromedial occipital region. Patients with 'spelling dyslexia' had large lesions of the ventral temporal lobe involving cortical regions believed to participate in later stages of visual processing. These findings provide a framework for the prediction of dyslexia type and severity based on lesion topography.

INTRODUCTION

In an early description of pure alexia, Dejerine (1892) used the term cecite verbale ('verbal blindness') to characterize the absolute inability of his patient to read words and letters aloud. Dejerine proposed that his patient's visual and language centres, although intact, had been disconnected by a lesion in the white matter pathway linking them. While a few similarly affected cases have been described by others (Caplan and Hedley-Whyte, 1974; Levin and Rose, 1979; Michel et al., 1979; Mori et al., 1982; Beauvois and Saillant, 1985; Lindeboom and Swinkels, 1986; Coslett and Saffran, 1989), it is clear that most patients with posterior dominant hemisphere injury do not show this same global alexia. Injury in the territory of the dominant posterior cerebral artery may produce no reading disorder (De Renzi et al., 1987), ill-defined deficits such as easy fatigue during reading (Castro-Caldas and Salgado, 1984), hemiparalexic disorders in which one end of words are misread (Binder et al., 1992), a 'spelling dyslexia' in which patients read aloud using a letter-by-letter reading strategy (Patterson and Kay, 1982) or a global alexia with subsequent recovery (Kurachi et al., 1979). These incomplete syndromes have been associated with a striking array of lesions of different size and location, many quite unlike the infarct of Dejerine's case. To complicate matters further, neuropsychological studies of spelling dyslexia suggest that it may be a qualitatively distinct syndrome, involving a whole-word recognition defect at a later processing stage than is Correspondence to: J. R. Binder, MD, Department of Neurology, The Medical College of Wisconsin, 9200 West Wisconsin Avenue, Milwaukee, Wl 53226, USA. © Oxford University Press 1992

Downloaded from http://brain.oxfordjournals.org/ by guest on April 12, 2016

SUMMARY

1808

J. R. BINDER AND J. P. MOHR

METHODS Patients The five global alexics included all right-handed patients known to us with dominant posterior cerebral artery territory infarction who met study criteria. Criteria consisted of an inability to read single words

Downloaded from http://brain.oxfordjournals.org/ by guest on April 12, 2016

assumed to be affected in global alexia (Warrington and Shallice, 1980; Friedman, 1982; Patterson and Kay, 1982). Spelling dyslexics typically have normal letter naming and rely on this capability for reading words and text. Global alexics, in contrast, show a deficit of letter naming that precludes attempts to read words aloud. Although studies subsequent to Dejerine's have clarified and elaborated upon his visualverbal pathway model (Geschwind, 1965; Damasio and Damasio, 1983; Henderson, 1986), none has offered an explanation for the existence of incomplete or qualitatively distinct syndromes. Dejerine's conception of the visual-verbal pathway as a compact white matter tract (Dejerine and Vialet, 1893) is particularly problematic in this regard, since injury to such a structure might not be predicted to result in highly variable reading deficits. An alternative possibility is that neuronal structures utilized during reading are not localized in a small anatomical area, and therefore not easily obliterated by a single lesion (Binder et ai, 1989). Recent evidence supporting a parallel distributed network organization of visual processing are consistent with such a suggestion (Ungerleider and Mishkin, 1982), as are anatomical findings that the occipito-temporal association 'pathway' consists of a system of multiple parallel cortical relays that have a wide distribution over the lateral and ventral brain surface (Rockland and Pandya, 1981; Tusa and Ungerleider, 1985). It should be possible, through an analysis of the dominant posterior cerebral artery lesions associated with different dyslexia syndromes, to specify the lesion characteristics (e.g. topography, volume) associated with each syndrome. The fact that reading may be normal after dominant posterior cerebral artery injury suggests that some posterior brain regions (e.g. primary visual cortex) are not critical for reading. Conversely, the lesions in cases with global alexia must include, for practical purposes, all of the pathways subserving reading in the posterior dominant hemisphere. Injury to a portion of these critical pathways might be expected to result in a partial syndrome such as spelling dyslexia. From such information an anatomical model of the pathways involved in reading could be elaborated to explain the observed diversity of clinical disorders. Such a model might predict the severity and general type of deficit to be expected from a given lesionpredictions not easily made from currently popular anatomical models of pure alexia. The present study concerns three groups of patients with dominant posterior cerebral artery territory infarction: (i) global alexics, (ii) spelling dyslexics ('verbal alexics', 'letter-by-letter readers', 'word-form dyslexics'), and (iii) normal readers. In our studies, these have been the general syndromes most commonly observed among dominant posterior cerebral artery stroke patients. We carried out a tomographic analysis of the lesion topography associated with each group using template overlap techniques. Lesions in the two groups of dyslexic patients were contrasted with those in the 'control' patients with dominant posterior cerebral artery infarction in whom reading was unaffected. Comparisons of mean total lesion volume and mean lesion area on each standard tomographic slice were carried out to assess volumetric effects on clinical outcome.

CALLOSAL READING PATHWAYS

1809

or text aloud and a severe deficit of letter naming (see below Reading tests) persisting for the duration of follow-up (at least 3 mths after stroke in all cases). All were encountered prospectively during the years 1987-1991. Age range was from 58 yrs to 82 yrs, with a mean of 71 yrs. There were three women and two men. The five spelling dyslexics included all right-handed patients known to us with dominant posterior cerebral artery territory infarction who met study criteria. Criteria consisted of rapid improvement in reading during the first months after stroke, including at least 90% accuracy in naming single letters, and development within the first month of letter-by-letter reading of words and text. All were encountered prospectively during the years 1987—1991. Age range was from 54 yrs to 80 yrs, with a mean of 72 yrs. All were men. The seven normal readers included all right-handed patients known to us with dominant posterior cerebral artery infarction who met criteria consisting of absence of reading complaints, normal letter naming, normal word reading and normal sentence reading when examined in the acute phase after stroke onset. All were encountered prospectively during the years 1987—1991. Age range was from 33 yrs to 68 yrs, with a mean of 51 yrs. There were four women and three men.

Anatomic analysis Lesion analysis was performed using computerized tomography (CT) scan data in all cases. All scans were obtained at the end of the first week after stroke onset, and all were taken in an axial plane ranging from approximately 10— 15 degrees above the canthomeatal line. Slice thickness was 7 mm in all cases. Scan slices were matched to reference 15-degree sections from the atlas of Matsui and Hirano (1978), and lesions were traced on computer templates constructed from these reference sections. Composite images of the lesions in each patient group were produced by digitally overlapping the individual tracings (pixel values arithmetically summed and then readjusted to a 0—256 scale) using a grey-scale image-processing program (see below, Figs 2—4). To clarify differences in lesion topography between abnormal and normal readers, the areas involved on the normal readers' composite were subtracted (pixel values arithmetically subtracted and then readjusted to a 0—256 scale) from the composites of the two abnormal groups to produce a global alexia-normal composite and a spelling dyslexia-normal composite. Area and volumetric measurements were made using the digitally templated lesion tracings on the 10 most frequently involved tomographic levels. Lesion area (number of darkened pixels) was calculated for each template level and divided by the total left hemisphere area at that level, to produce a lesion area index representing the percentage of the left hemisphere area involved by the lesion. The sum of the lesion areas (all darkened pixels) on the 10 scan levels was then divided by the sum of the left hemisphere areas on the 10 scan levels to produce a lesion volume index. The lesion volume index represents the percentage of the total left hemisphere volume involved by the lesion over the 10 scan levels analysed. Mean lesion area index for each template level, and mean lesion volume index, were calculated for the three clinical groups. Between-group differences were tested for significance using paired two-tailed / tests.

Downloaded from http://brain.oxfordjournals.org/ by guest on April 12, 2016

Reading tests Test items included: (i) the 26 letters of the alphabet, printed as single capitals 1 inch high, on 2.5x2.5-inch white paper squares, which the subject was required to name aloud; (ii) a set of 292 single words varying in length, frequency and concreteness, which included sublists of 50 grammatical 'functor' words and 42 pronounceable non-words, printed individually in lower-case large print (height 1 inch) on white paper, which the subject was required to read aloud; (iii) paragraph items from the Gray Oral Reading Test (Gray, 1955), which were read aloud and timed. Word reading was begun using short, high-frequency nouns and discontinued if completely unsuccessful after 10 — 15 items, so as not to jeopardize patient cooperation or cause extreme distress. Fluent readers were given much larger numbers of items, in an effort to detect mild deficiencies. Only the first complete word response was scored in assessing accuracy. A representative reading rate for easy sentence material was derived by averaging rates obtained on paragraphs 1 and 4 of the Gray Oral Reading Test. All patients with abnormal performances were given follow-up testing to measure the extent and longitudinal profile of recovery. Patients unable to read aloud were tested for printed word comprehension using a written adaptation of the Peabody Picture Vocabulary Test (Warrington and Shallice, 1979; Dunn and Dunn, 1981). In addition to the reading battery, all patients had formal visual field testing by either tangent screen or automated perimetry.

1810

J. R. BINDER AND J. P. MOHR

RESULTS

Reading data Best oral reading performances and visual field results for the 17 patients are summarized in Table 1. None of the five global alexics (Table 1, nos 1 —5) succeeded

TABLE I. PATIENT DATA Visual fields RHH RHH RHH RHH RHH

Time tested 20 mths 26 mths 5 mths 3 mths 16 mths

Sentence reading Unable Unable Unable Unable Unable

Word reading2

The topography of callosal reading pathways. A case-control analysis.

Lesion topography and reading ability were analysed in 17 patients with dominant posterior cerebral artery territory infarction. Patients with dominan...
5MB Sizes 0 Downloads 0 Views