Novel Methods to Predict Outcome Using Neuroimaging

September 1, 2006
Alexander Gantman, MA

Volume 23, Issue 9

Noninvasive brain imaging methods are providing unprecedented views of the structural and functional development and aging of an individual's brain or state of brain pathology. These exciting new may provide novel information relevant to the enhancement of clinical practice.

The capacity of cognitive neuroscience to inform clinical practice has stimulated both excitement and controversy.1-5 Noninvasive brain imaging methods are providing unprecedented views of the structural and functional development and aging of an individual's brain or state of brain pathology. These exciting new views of maturing and aging brains or of brain pathology may provide novel information relevant to the enhancement of clinical practice. The implications of neuroscience for clinical practice, however, have raised some controversy because of speculative and potentially flawed interpretations created by associating animal experimentation with human medical practice.1 As a result, the relationship between clinical practice and neuroscience has been called "a bridge too far."6 Nonetheless, it is evident that the field of cognitive neuroscience holds promise to inform clinical practice. The promise lies in its potential to address critical issues such as:

  • Understanding the neural processes underlying the development of essential skills, such as reading and numeracy, and the "healthy" decline of cognitive functions such as memory.
  • Understanding the pathophysiology of various neuropsychiatric disorders.
  • Assessment and prediction of interventional outcome.
  • Early identification of persons at risk for neuropsychiatric disorders.

While the following discussion focuses mainly on reading skills and disabilities, similar arguments may be applied to clinical issues. We focus on recent findings from our laboratory that highlight the extent to which functional and structural neuroimaging techniques can serve as a clinically useful tool in predicting outcome.

Reading is a skill that is undeniably critical to success in modern society, yet the development of reading abilities in children is becoming a significant problem. Based on the National Assessment of Educational Progress,7 only 31% of the nation's fourth graders are performing at or above the proficient achievement level that demonstrates solid academic performance in reading. Furthermore, dyslexia, a developmental learning disability that is characterized by difficulty in reading in individuals who otherwise have the intelligence and necessary motivation for accurate and fluent reading,8 is prevalent in 5% to 17% of all children and 80% of children with learning disabilities.9

Reading First, an initiative under the No Child Left Behind Act of 2002, was enacted with the goal of achieving high reading proficiency by the end of the third grade by 2014. The initiative emphasizes early identification of children who are at risk for reading failure. Once they are identified, neuroscientifically proven reading programs, interventions, and strategies in the early grades are used to improve reading fluency.

Neuroscientific evidence

Neuroscientific evidence from a number of brain imaging studies performed over the past decade has provided us with novel insights into our knowledge of normal and disturbed reading.10-13 These studies have demonstrated that reading activates a widely distributed set of areas in the occipitotemporal; posterior temporal to parietotemporal; precentral; and inferior frontal gyri, which sustain orthographic, semantic, and phonologic processing.14 It has also been shown that the left posterior superior temporal, parietotemporal, and occipitotemporal regions are dysfunctional--with abnormal increases in activation in the left frontal region--in dyslexic readers.15-18 These findings have implications for novel models of reading and the disabilities associated with it.

Recent studies have shown the effects of interventions on neural patterns in response to reading.19-22 These studies demonstrate normalization of brain activation in the left hemispheric regions, which is critical for reading and generally dysfunctional in persons with dyslexia. In addition, an increased activation in the homologous right hemispheric regions is found in dyslexic brains after intervention. These studies provide insights into plastic changes that could occur in regions related to normal reading as well as putative compensatory changes in response to successful intervention. If examined carefully and with a larger number of subjects, these studies have the potential to assign optimal intervention strategies to children with different behavioral and neural profiles.

Investigation of the extent to which neuroimaging techniques can serve as a clinically useful tool in predicting future reading skills is still in its infancy. There are only a few developmental studies predicting reading skills and language, all of which use event-related potentials.23-25 Another study examining brain morphology using voxel-based morphometry (VBM) predicted short-term learning of novel speech sounds in adults.26 While there have been no studies that used functional neuroimaging to predict outcome in reading, there are an increasing number of studies with functional imaging in disorders such as depression27-29 or Alzheimer disease.30-33

It is difficult to discern the clinical use of neuroimaging techniques in predicting outcome in these studies for the following reasons. First, it is unclear whether neuroimaging can predict outcome significantly better than other existing measures, such as behavioral testing (currently the most straightforward and inexpensive way of predicting reading outcome). In order to make the results clinically useful, it is important that neuroimaging techniques exceed currently available methods or have an add-on effect. Second, no validation or reliability analyses have been performed (except by Apostolova and colleagues,33 who used permutation tests). Finally, prospective analyses have not been performed; rather, all studies thus far have relied on retrospective correlational analyses of behavioral outcome and initial neuroim-aging measures.

However, one intriguing study on predicting memory decline in patients with postoperative temporal lobe epilepsy (TLE) has overcome many of these limitations.34 The authors examined preoperative behavior, brain volume, and functional MRI (fMRI) in a small sample of 10 patients with TLE to predict postoperative memory. They found that left-right hippocampal encoding activity difference showed reasonable sensitivity, specificity, and positive predictive value (20% to 100%) for predicting the amount of preoperative to postoperative memory decline.

Building upon these earlier studies, we addressed many of the limitations by comparing neuroimaging with existing assessment methods, performing validation analyses, and performing prospective analyses in a large sample of subjects. Our primary goals were to test whether neuroimaging can be used to predict reading success and to test whether we can achieve greater predictability by integrating neuroimaging measures into the currently available method of behavioral measures to predict reading success. We considered prediction to be an important goal because improved prediction of reading skills can facilitate identification of children who may benefit most from intensified or alternative reading instruction so that reading failure is minimized or even prevented.

Our present study focused primarily on one reading skill as an outcome measure that is thought to be essential for effective reading: word decoding. Decoding refers to the ability to determine the sound of a word from letters and syllables. Decoding ability is fundamental to reading because learning to read involves learning to relate the sounds of known auditory language (phonology) to letters (orthography). It is known that early and systemat- ic emphasis on decoding leads to superior reading skills,35,36 that decoding accounts for most of the variance in reading comprehension,37-39 and that the development of language-specific phonology is essential for reading success.40 Therefore, better methods for early identification of young children at risk for impaired decoding abilities hold promise for improving the specificity and effectiveness of early intervention and later achievement of reading skills.

A relatively pure test of word decoding involves reading pronounceable nonsense words aloud, because their proper pronunciation can only be derived from decoding skills (as opposed to words memorized by sight and context). Such a test also measures phonemic awareness, which is the awareness that separable sounds (phonemes) are blended to produce words. Phonemic awareness has been found to be one of the best predictors of reading success.41-45

Findings from a retrospective study

In our first study, we examined 53 normal readers aged 8 through 11 years, to find how behavioral and brain measures taken in the autumn of an academic year (Time 1) predicted decoding skills in the spring of the same academic year (Time 2) by measuring performance on the Woodcock Reading Mastery Test's (WRMT) Word Attack (WA) subtest.46 WA requires the child to attempt to read pronounceable nonwords of successive difficulty aloud. Initial assessment was performed using a standard set of reading and other behavioral tests, fMRI during a real-word rhyme judgment task (a standard fMRI task that taps into phonemic awareness, as does WA), and VBM measures of gray- and white-matter densities.

Using whole-brain regression analyses, we identified regions that showed significant correlation with brain activation (P < .001, uncorrected) or gray- or white-matter volume (P < .001, family-wise-error corrected) at Time 1 that correlated with Time 2 WA standard scores (WA-ss) adjusted for age. We then extracted contrast estimates or volume measures from these regions and submitted them to multiple regression analysis.

We found that specific patterns of brain activation during phonologic processing and gray- and white-matter morphology in regions critical for normal reading were correlated with later reading skills (Figure 1A) and that the neuroimaging model explained 66% of the variance (Figure 1B). We also identified behavioral tests at Time 1 that correlated with Time 2 WA-ss and performed similar multiple regression analyses. In contrast, the behavioral model, which consisted primarily of measures related to phonemic awareness, explained only 49% of the variance (Figure 1C).

Interestingly, a model combining behavioral and neuroimaging measures predicted future reading abilities significantly better than the behavioral or neuroimaging model alone, explaining 78% of the variance (Figure 1D). Various validation analyses--including leave-one-out cross-validation analysis--showed that the combined model of behavioral and neuroimaging measures showed significantly less deviation between the actual outcome scores and predicted scores, compared with the behavioral or neuroimaging models (both, P < .05; Figure 1E). There were no significant differences between the behavioral and neuroimaging models (P > .05). We further replicated the basic findings using a larger sample of 64 children who had variable reading ability ranging from poor to good (F. Hoeft, MD, unpublished data, 2006). These studies showed that neuroimaging measures can predict future reading ability; combined with behavioral measures, neuroimaging can be a powerful way to predict future reading ability.

Findings from a prospective study

Based on these promising results derived from a retrospective prediction model, the next study sought to prospectively predict children's future reading skills.47 Thus, this study went beyond correlating behavioral outcome with initial behavioral and neuroimaging data. The sample consisted of 59 children aged 8 through 12 years at the time of initial assessment (Time 1). The children included both good and poor readers. WA-ss and neuroimaging measures collected at Time 1 and WA-ss collected approximately 6 months later (Time 2) were used to predict reading ability as of approximately 2 years later (ie, WA-ss at Time 3). The neuroimaging model created was similar to that of the previous studies in our laboratory. For some analyses, we also used reading comprehension scores (WRMT Passage Comprehension standard scores [PC-ss]) as the outcome measure.

The results of this study indicated that the neuroimaging model was as successful as Time 1 WA-ss was in predicting future reading ability (r = 0.55 and r = 0.57, respectively; both, P < .001). We further assessed the clinical use of determining reading disability (RD) by examining sensitivity, specificity, and positive predictive values. Various criteria to define RD (WA-ss < 85, 86, 87, 88, 89, or 90) were tested. Results from this analysis indicated that the combination of behavioral and neuroimaging measures was now superior to using Time 1 WA-ss (Figure 2A). It is thought that the sensitivity index, specificity index, and positive predictive values should all reach at least 0.75 in order for a measure to be considered acceptable for practical use and suitable for screening purposes.48 It is promising that the combination of existing behavioral measures and novel neuroimaging measures passed these criteria and outperformed behavioral measures. Finally, when we applied these models to predict reading comprehension skills (PC-ss)--which is the ultimate reading goal--similarly high sensitivity, specificity, and positive predictive value were achieved (Figure 2B).

Limitations and future direction

While we found that combining behavioral and neuroimaging measures can predict future reading ability significantly better than using the behavioral or neuroimaging models alone and that prospective analyses showed promise for practical use, the results thus far have several limitations. First, the patients were followed up for only 1 or 2 school years. (It will be interesting to examine the predictive value in the long term.) Second, the behavioral and neuroimaging predictors used here were selected from multivariate analyses from the total sample and were applied to the validation tests. Finally, the models examined linear relationships, and an increasing number of studies show nonlinear effects of development.49 These possible limitations may have potentially biased the data, which should be interpreted with caution until results are replicated in larger samples. Future studies of preschool children before they learn to read are warranted and should include environmental variables, such as socioeconomic status, and genetic measures, which play a large part in development of reading disabilities.

Conclusion

Our results raise the intriguing possibility of using neuroimaging data as a critical component in the assessment and prediction of children's reading skills. If we are able to overcome the aforementioned limitations, our results may have a significant impact on public health, since it may help identify at-risk children at an early stage and provide them with opportunities to receive optimal interventions earlier. Furthermore, these studies may help bridge the research and actual practice of psychiatrists, neurologists, and educators. Finally, similar models using neuroimaging or a combination of behavioral and neuroimaging measures may be used to predict the development or decline of other cognitive functions as well as therapeutic outcomes.

Mr Gantman and Ms Wittenberg are researchers in the department of psychiatry and behavioral sciences in the Center for Interdisciplinary Brain Sciences Research (CIBSR) at the Stanford University School of Medicine and members of the PGSP (Pacific Graduate School of Psychology)-Stanford PsyD Consortium in Palo Alto, Calif. Dr Hoeft is senior research scientist in the department of psychiatry and behavioral sciences in the CIBSR at the Stanford University School of Medicine. They report that they have no conflicts of interest concerning the subject matter of this article.

References:

References

1.

Bruer JT. Avoiding the pediatrician's error: how neuroscientists can help educators (and themselves).

Nat Neurosci.

2002;5(suppl):1031-1033.

2.

Goswami U. Neuroscience and education: from research to practice?

Nat Rev Neurosci

. 2006;7:406-411.

3.

Moats L. Relevance of neuroscience to effective education for students with reading and other learning disabilities.

J Child Neurol

. 2004;19:840-845.

4.

Koizumi H. The concept of 'developing the brain': a new natural science for learning and education.

Brain Dev

. 2004;26:434-441.

5.

Honey GD, Fletcher PC, Bullmore ET. Functional brain mapping of psychopathology.

J Neurol Neurosurg Psychiatry

. 2002;72:432-439.

6.

Bruer JT. Education and the brain: a bridge too far.

Ed Res

. 1997;26:1-13.

7.

Trial Urban District Assessment Reading 2005. National Assessment of Educational Progress. February 2006. Available at: http://nces.gov/nationsreportcard/pdf/dst2005/2006455r.pdf. Accessed August 8, 2006.

8.

Lyon GR. Research initiatives in learning disabilities: contributions from scientists supported by the National Institute of Child Health and Human Development.

J Child Neurol

. 1995;10(suppl 1):S120- S126.

9.

Shaywitz SE. Dyslexia.

N Engl J Med

. 1998;338: 307-312.

10.

Price CJ, Mechelli A. Reading and reading disturbance.

Curr Opin Neurobiol

. 2005;15:231-238.

11.

Demonet JF, Taylor MJ, Chaix Y. Developmental dyslexia.

Lancet

. 2004;363:1451-1460.

12.

Habib M. The neurological basis of developmental dyslexia: an overview and working hypothesis.

Brain

. 2000;123:2373-2399.

13.

Shaywitz SE, Shaywitz BA. Dyslexia (specific reading disability).

Biol Psychiatry

. 2005;57:1301-1309.

14.

Turkeltaub PE, Eden GF, Jones KM, Zeffiro TA. Meta-analysis of the functional neuroanatomy of single-word reading: method and validation.

Neuroimage

. 2002;16:765-780.

15.

Shaywitz BA, Shaywitz SE, Pugh KR, et al. Disruption of posterior brain systems for reading in children with developmental dyslexia.

Biol Psychiatry

. 2002;52: 101-110.

16.

Shaywitz SE, Shaywitz BA, Pugh KR, et al. Functional disruption in the organization of the brain for reading in dyslexia.

Proc Natl Acad Sci U S A

. 1998;95: 2636-2641.

17.

Temple E, Poldrack RA, Protopapas A, et al. Disruption of the neural response to rapid acoustic stimuli in dyslexia: evidence from functional MRI.

Proc Natl Acad Sci U S A

. 2000;97:13907-13912.

18.

Hoeft F, Hernandez A, McMillon G, et al. Neural basis of dyslexia: a comparison between dyslexic and non-dyslexic children equated for reading ability.

J Neuroscience

. In press.

19.

Aylward EH, Richards TL, Berninger VW, et al. Instructional treatment associated with changes in brain activation in children with dyslexia.

Neurology

. 2003; 61:212-219.

20.

Eden GF, Jones KM, Cappell K, et al. Neural changes following remediation in adult developmental dyslexia.

Neuron

. 2004;44:411-422.

21.

Shaywitz BA, Shaywitz SE, Blachman BA, et al. Development of left occipitotemporal systems for skilled reading in children after a phonologically-based intervention.

Biol Psychiatry

. 2004;55:926-933.

22.

Temple E, Deutsch GK, Poldrack RA, et al. Neural deficits in children with dyslexia ameliorated by behavioral remediation: evidence from functional MRI.

Proc Natl Acad Sci U S A

. 2003;100:2860-2865.

23.

Lyytinen H, Ahonen T, Eklund K, et al. Early development of children at familial risk for dyslexia--follow-up from birth to school age.

Dyslexia

. 2004;10: 146-178.

24.

Molfese VJ, Molfese DL, Modgline AA. Newborn and preschool predictors of second-grade reading scores: an evaluation of categorical and continuous scores.

J Learn Disabil

. 2001;34:545-554.

25.

Espy KA, Molfese DL, Molfese VJ, Modglin A. Development of auditory event-related potentials in young children and relations to word-level reading abilities at age 8 years.

Ann Dyslexia

. 2004;54:9-38.

26.

Golestani N, Paus T, Zatorre RJ. Anatomical correlates of learning novel speech sounds.

Neuron

. 2002; 35:997-1010.

27.

Canli T, Cooney RE, Goldin P, et al. Amygdala reactivity to emotional faces predicts improvement in major depression.

Neuroreport

. 2005;16:1267-1270.

28.

Davidson RJ, Irwin W, Anderle MJ, Kalin NH. The neural substrates of affective processing in depressed patients treated with venlafaxine.

Am J Psychiatry

. 2003; 160:64-75.

29.

Siegle GJ, Carter CS, Thase ME. Use of FMRI to predict recovery from unipolar depression with cognitive behavior therapy.

Am J Psychiatry

. 2006;163: 735-738.

30.

Bookheimer SY, Strojwas MH, Cohen MS, et al. Patterns of brain activation in people at risk for Alzheimer's disease.

N Engl J Med

. 2000;343:450-456.

31.

de Leon MJ, Convit A, Wolf OT, et al. Prediction of cognitive decline in normal elderly subjects with 2-[(18)F]fluoro-2-deoxy-D-glucose/positron-emission tomography (FDG/PET).

Proc Natl Acad Sci U S A

. 2001; 98:10966-10971.

32.

Mungas D, Harvey D, Reed BR, et al. Longitudinal volumetric MRI change and rate of cognitive decline.

Neurology

. 2005;65:565-571.

33.

Apostolova LG, Dutton RA, Dinov ID, et al. Conversion of mild cognitive impairment to Alzheimer disease predicted by hippocampal atrophy maps.

Arch Neurol

. 2006;63:693-699.

34.

Richardson MP, Strange BA, Thompson PJ, et al. Pre-operative verbal memory fMRI predicts post-operative memory decline after left temporal lobe resection.

Brain

. 2004;127:2419-2426.

35.

Adams MJ.

Beginning to Read: Thinking and Learning About Print

. Cambridge, Mass: MIT Press; 1990.

36.

Snow CE, Burns SM, Griffin P, eds.

Preventing Reading Difficulties in Young Children

. Washington, DC: National Academy Press; 1998.

37.

Ehri LC, Robbins C. Beginners need some decoding skills to read by analogy.

Reading Res Quart

. 1992; 27:13-26.

38.

Gough PB, Tunmer WE. Decoding, reading and reading disability.

Remedial Special Ed

. 1986;7:6-10.

39.

Richardson E, DiBenedetto B, Adler A. Use of the decoding skills test to study differences between good and poor readers.

Adv Learning Behav Disabil

. 1982; 1:25-74.

40.

Snowling MJ.

Dyslexia: A Cognitive Developmental Perspective

. Oxford, UK: Basil Blackwell; 1987.

41.

Bradley L, Bryant PE. Categorizing sounds and learning to read: a causal connection.

Nature

. 1983;30: 419-421.

42.

Juel C. Learning to read and write: a longitudinal study of 54 children from first through fourth grades.

J Ed Psychol

. 1988;80:437-447.

43.

Lundberg I, Olofsson A, Wall S. Reading and spelling skills in the first school years predicted from phonemic awareness skills in kindergarten.

Scand J Psychol

. 1980;21:159-173.

44.

Muter V, Hulme C, Snowling M, Taylor S. Segmentation, not rhyming, predicts early progress in leaming to read.

J Child Psychol Psychiatry.

1997;35: 293-310.

45.

Naslund JC, Schneider W. Kindergarten letter knowledge, phonological skills, and memory processes: relative effects on early literacy.

J Exp Child Psychol

. 1996; 62:30-59.

46.

Hoeft F, Ueno T, Hernandez A, et al. Predicting children's later decoding skills using behavioral, functional and structual neuroimaging measures.

J Cog Neurosci

. 2006(suppl):P206.

47.

Kobayashi N, Meyler A, Keller TA, et al. Neuroimaging measures can prospectively predict future reading skills. Poster to be presented at: Neuroscience 2006; October 14-18, 2006; Atlanta.

48.

Gredler GR. Early childhood screening for developmental and educational problems. In: Bracken BA, ed.

The Psychoeducational Assessment of Preschool Children

. Boston: Allyn & Bacon; 2000:399-411.

49.

Shaw P, Greenstein D, Lerch J, et al. Intellectual ability and cortical development in children and adolescents.

Nature

. 2006;440:676-679.