The following is a guest post by Whitney Anne Postman-Caucheteux.
If there were a sub-field of Aphasiology devoted to imaging of the brains of people with aphasia, let’s call it “Neuro-Aphasiology”, then Dr. Julius Fridriksson would be one of its most distinguished pioneers. As a long-time admirer of his work, I can think of few other researchers of aphasia who have gone beyond simply talking about issues such as age, task difficulty, and perfusion in neuroimaging of language processes in people with aphasia, to actually conducting and publishing the foundational research (see Fridriksson et al 2006, 2005, 2002, among others).
With two recently published aphasia fMRI papers, Dr. Fridriksson and his team at the University of South Carolina have done it again, by combining advanced fMRI techniques for acquiring overt speech responses with sophisticated psycholinguistic analyses of word production in aphasia. I propose that these two papers should be read as a pair, since each provides complementary investigations of the contributions of perilesional and contralesional regions to language production in chronic aphasia:
“F1”: Fridriksson, J., Baker, J.M. & Moser, D. (2009). Cortical mapping of naming errors in aphasia. Human Brain Mapping, 30, 2487-2498.
“F2”: Fridriksson, J., Bonilha, L., Baker, J.M., Moser, D., & Rorden, C. (in press). Activity in preserved left hemisphere regions predicts anomia severity in aphasia. Cerebral Cortex.
Both papers (hence, “F1” and “F2”) describe Fridriksson et al’s overt picture-naming experiments with chronic stroke patients with aphasia using fMRI. F1 is perhaps the more revolutionary of the two, in being the first to link certain patterns of neural activation in such patients with specific error types that have long been the subject of psycholinguistic investigations (e.g., Schwartz et al, 2006). I will review F1 and F2 in turn before offering suggestions on how they complement each other in providing clues to different pieces of the puzzle of language production in post-stroke aphasia.
Part I on F1: Fridriksson, J., Baker, J.M. & Moser, D. (2009). Cortical mapping of naming errors in aphasia. Human Brain Mapping, 30, 2487-2498.
In F1, Fridriksson et al employed a sparse sampling technique to acquire overt naming responses to object pictures from 11 stroke patients with various types of aphasia and with a range of degrees of anomia severity, all in chronic stages. Their goal was to identify common areas of activation across the entire cohort associated with 1) accurate picture-naming, 2) phonemic errors, and 3) semantic errors. This goal is crucial to understanding the neural substrate of disordered language production post-stroke, and a valuable use of novel techniques for acquisition of overt speech with fMRI. Dr. Bruce Crosson and colleagues have elegantly outlined their recommendations for how best to acquire, analyze and interpret fMRI data from language production tasks with patients with aphasia in Crosson et al (2007). In addition to the familiar complications of having stroke patients with aphasia participate in fMRI studies, acquisition of overt speech responses from such patients during scanning can be confounded by related motor-speech disorders such as apraxia, and the possibility of extremely long response times for some patients.
Prior fMRI research using silent production could not distinguish between activation patterns associated with accurate and inaccurate responses. Distinctive patterns are to be expected, given results from studies linking superior language production performance (measured outside the scanner with fMRI, during scanning with PET) with predominantly perilesional activation, and inferior performance with increased contralesional involvement (see references in F1 and F2). Consequently, studies like F1 are needed to discover how neural patterns may differ for accurate and inaccurate naming. Such discoveries can clarify effective vs. ineffective ways in which neural systems respond to damage, and subsequently, how these ways can be enhanced or suppressed with treatment.
Focusing on areas of activation common to the cohort, Fridriksson et al masked out voxels lesioned in any of the 11 patients, extending over the greater portion of the left hemisphere. The patients achieved a wide range of accuracy, and committed semantic errors, phonemic errors, unrelated errors, neologisms or omissions (see Table II in F1 reproduced and modified below).
The authors correlated the patients’ correct responses and semantic and phonemic errors with increases in BOLD activation in neural regions outside of the aforementioned mask, yielding the following intriguing results:
Result #1: Correct names correlated positively with increases in BOLD response in right inferior frontal gyrus.
Result #2: Phonemic errors correlated positively with increases in BOLD response in left precuneus (BA19), cuneus (BA7) and posterior inferior temporal gyrus (BA37).
Result #3: Semantic errors correlated positively with increases in BOLD response in right cuneus (BA 18), middle occipital gyrus (BA 18/19), and posterior inferior temporal gyrus (BA37).
The first result linking naming accuracy to activation in the right IFG is corroborated by some previous research suggesting positive contributions of this region to successful language production (see references in F1). It is also ostensibly in contradiction with the principal results of F2, but more on that issue in Part III. Here I’d like to concentrate on the top graph in Figure 3:
The tight correlation between accuracy and right IFG activation is striking. Yet even so, it is worth mentioning that for 4 patients, virtually 0% or less than .1% increase in BOLD amplitude was coupled with naming accuracy, raising the question of whether this result was carried largely by a subset of the cohort. If this were indeed the case, it might help to elucidate which patients are expected to show substantial right IFG activation linked to accuracy. This issue was raised in Postman-Caucheteux et al (in press), in a discussion of important case studies by Meinzer et al (2006) and Vitali et al (2007).
The novel results coupling phonemic errors with ipsilesional posterior activation, and semantic errors with contralesional posterior activation, should inspire future research directed at replicating and developing them further. The explanations offered by the authors for why these regions should be involved in the production of these errors are plausible and appealing. Especially interesting was their finding that the neural activation patterns linked to each of these patterns were essentially additional to that observed for correct naming. That is, they both involved the same neural substrate as accuracy, plus activation in the aforementioned posterior areas. This finding is in agreement with that for incorrect vs. correct naming in Postman-Caucheteux et al (in press), although we found a link between right frontal, not posterior, regions for semantic paraphasias (as well as omissions). However, the patients in our smaller cohort had frontal-insular-parietal damage, with almost no temporal damage. Since the brain region most affected in the F1 cohort was posterior temporal, this comparison raises the possibility that semantic errors may principally involve directly contralesional activation, i.e., activation in right frontal regions in patients with left frontal lesions, and in right posterior regions in patients with left posterior lesions. A worthwhile approach to investigate this possibility would take into account the precise nature of the semantic errors, which brings me to my next point.
More qualitative details (including examples) from all of the error types, and measurement of reaction times as an index of naming difficulty, would have been informative. I would also like to know if the high number of unrelated errors produced by P3 were perseverations, which may constitute their own special class of errors. Likewise, more information on the types of semantic errors could have been used to support the authors’ interpretation of right posterior activation as representing less specific semantic representations (p.2496). Furthermore, research on the evolution of neologisms into phonemic paraphasias (Bose & Buchanan, 2007) implies that comparison of possible neural patterns for neologisms with those found for phonemic errors could have been instructive.
The grounds for the authors’ exclusion of other types of errors in their analyses are somewhat unclear, for even though phonemic and semantic errors were the most frequent types, the other types of errors were not infrequently produced by certain patients. With regard to omissions, even though the interpretation of omissions is indeed problematical, nevertheless they are routinely tracked as errors, and they can be predicted by specific psycholinguistic factors such as semantic competition (Schnur et al, 2006). Since the authors did not include an analysis of factors that could have contributed to each type of error (e.g., percent name agreement, age of acquisition, target word length), it is unknown whether certain stimuli were consistently more likely to induce errors, as was found in Postman-Caucheteux et al (in press). Given that the effects of different psycholinguistic variables on picture-naming have been linked to specified neural areas of activation (Schnur et al 2009, Wilson et al 2009), future studies inspired by F1 should seek to isolate the variables that induce errors, perhaps by manipulating stimuli according to factors of interest and measuring reaction times. This approach may be helpful for interpreting the nature of error-linked activation.
As corroboration of their findings in F1, the authors cite a clever treatment study of word learning using PET by Raboyeau et al (2008). Chronic stroke patients with aphasia were trained to produce names of objects that had been difficult for them prior to therapy. At the same time, healthy participants were trained to produce words in second languages that they had acquired in school with varying degrees of proficiency. Raboyeau et al’s findings of increased right insular and frontal activation with word learning in both groups are interpreted in F1 along these lines:
“[They] concluded that increased activity in the right frontal lobe in aphasia is not merely the consequence of damaged homologues in the left hemisphere but, rather, is a reflection of increased reliance on the right hemisphere to support aphasia recovery” (p. 2496).
Raboyeau et al’s findings may actually be trickier to explain, as they also included more activation in left frontal regions (BA’s 10 and 11) in the patients but not the controls. Additional left hemisphere activation may have been present in some patients but, as with F1, lesioned voxels were excluded from their analyses. Here is how Raboyeau et al state their own conclusions:
“[...] Activations observed in these two right frontal regions do not seem to play a true compensatory function in aphasia (italics mine), and do not represent a mere consequence of left hemispheric lesion, as they existed also in non–brain-damaged subjects [...]” (p.296).
So as I understand their discussion, they did not infer that right insular and frontal activation supported recovery, as suggested in F1. Rather, their results indicated greater effort and cued word retrieval as a result of training, in patients as well as controls. If I have misconstrued Raboyeau et al’s and F1’s conclusions, hopefully someone will help me see the light.
Part II on F2: Fridriksson, J., Bonilha, L., Baker, J.M., Moser, D., & Rorden, C. (in press). Activity in preserved left hemisphere regions predicts anomia severity in aphasia. Cerebral Cortex.
While the fMRI task and methods of acquisition for F2 appear to be almost identical to those in F1, the leading question and analytical techniques were virtually the converse. Instead of focusing on patients’ errors as in F1, here the authors asked which brain regions appear to support accurate overt picture naming. Also, instead of analyzing the cohort of patients as a group and creating a group lesion mask, here the authors examined the patients (N=15) individually and compared each one’s activation map to the average from an equal number of healthy age-matched control participants.
Since here I could not do justice to their advanced methods, readers are referred to the original paper for details on the complex steps involved in comparing activation maps derived from each patient’s contrast of correct picture naming vs. abstract (silent) picture viewing, with the controls’ group map. In essence, the degree to which each patient’s activation map deviated from the average control map was correlated with their proportion of correct naming responses. In addition, structural analyses were conducted to investigate whether intensity of activation associated with correct naming was dependent upon specific areas of damage. Results were:
1) In the control group, picture-naming was supported by bilateral activation in posterior regions (cuneus & inferior/middle occipital gyrus (BA 18), middle temporal gyrus (BA37)) but highly left lateralized in the transverse (BA 42) and superior (BA 22) temporal gyri, and frontally in inferior frontal gyrus (BA 45), middle frontal gyrus (BA 10, 11, 47), and anterior cingulate (BA 32).
2) For the 15 participants with aphasia due to left-hemisphere stroke, accurate picture naming was supported by many of the same left-lateralized regions observed in the control group. Most of these were perilesional, namely, medial & middle frontal gyrus (BA 10, 11, 47) and inferior occipital gyrus (18). The left anterior cingulate gyrus (BA 32) was also linked to correct naming in patients, but was considered too medial to qualify as perilesional for this cohort of patients.
3) In the patients, intensity of activation in these left-lateralized areas correlated with number of correct names. Here’s the money shot (Figure 3 in F2), showing cortical areas associated with naming task performance (red-yellow scale) along with the lesion overlay map for all 15 patients (blue-green scale):
4) But Fridriksson et al didn’t stop there. Even more dazzling, those patients who did best on the naming task in the scanner tended to show greater activation than the controls in the regions highlighted above (red-yellow), and those who did less well on the naming task showed less activation than the controls in the same areas. Figure 4 in F2 is copied below, showing “the relationship between intensity of activation (x-axis; measured in Z-scores compared with a group of normal control participants) and the number of correct naming attempts (y-axis; out of 80 pictures) during fMRI scanning”:
5) A final intriguing result: Intensity of activation in the patients was inversely correlated with damage to the posterior left IFG (BA 44).
The findings in 2) are corroborated by those in Postman-Caucheteux et al (in press), showing predominantly left perilesional activation for accurate picture-naming in patients with frontal-insular-parietal damage. To my knowledge, the findings in 3) and 4) provide the most precise characterization of ipsilesional (including perilesional and non-perilesional) areas of activation for language production in aphasia, and the most direct link between this activation and production performance, yet to be discovered.
Part 3: Sum Total of F1 + F2
F2 contributes to the mounting evidence from the nascent wave of overt speech fMRI studies, for the fundamental importance of restoration or re-integration of perilesional tissue for good language production in people with chronic aphasia due to stroke (see references in Fridriksson et al (in press) and Postman-Caucheteux et al (in press)). So I couldn’t help but wonder, how would the authors relate these findings to their earlier paper (F1), in which they found a positive role for the right IFG in patients’ accurate naming? As I understand them, the results of F1 and F2 raise the following possibilities:
1) Contralesional (right) IFG may be working in tandem with perilesional regions (and perhaps also ipsilesional non-perilesional regions such as anterior cingulate) to achieve accurate naming in some patients. Thus in F1, some patients who showed increasing right IFG involvement with increasing naming success could have also shown substantial perilesional involvement that was not observable due to the group lesion mask. If this were the case, then it would provide evidence for partnership, rather than competition, between frontal areas of both stroke and non-stroke hemispheres.
2) In some patients, contralesional activation may be so negligible in comparison with robust perilesional activation that it only becomes apparent when large portions of the left hemisphere are masked out by group lesion analyses, as in F1. Presumably, some of the patients in F2 could have also shown right IFG involvement in successful naming, but the intensity may have been too minimal relative to ipsilesional areas to be reliably detected.
I’d like to propose that contralesional IFG activation might be helpful for good language production jointly with ipsilesional areas and up to a certain relatively low threshold. When it exceeds this threshold, it might constitute over-activation that is not effective, may be more evident for naming errors (as observed in Postman-Caucheteux et al, in press), and may even interfere with functioning of ipsilesional areas (Martin et al, 2009).
In the two studies reviewed here, Dr. Fridriksson’s team found contributions of perilesional and contralesional activation to language production in post-stroke aphasia. A major step forward in disentangling these contributions has been achieved with their identification of areas involved in certain types of naming errors, signaling the way for future fMRI studies to appreciate the details of patients’ production performance. Moreover, they have deepened our understanding of the tight link between activation in certain ipsilesional areas and successful overt word production. To continue progressing in the direction led by Fridriksson et al, recognition of functional partnership between stroke and non-stroke hemispheres, and distinction between effective activation and ineffective/maladaptive over-activation of contralesional areas, may be helpful in future investigations and discussions.
1. The row indicating categories of nonfluent and fluent participants was added here. It does not appear in the original Table II in F1, p.2492.
2. The statistical methods employed by Fridriksson et al were much more sophisticated than mere correlation, but they will not be described in depth here.
Bose, A., & Buchanan, L. (2007). A cognitive and psycholinguistic investigation of neologisms. Aphasiology, 21, 726-738.
Crosson, B., McGregor, K., Gopinath, K.S., Conway, T.W., Benjamin, M., Chang, Y.L., et al. (2007). Functional MRI of language in aphasia: A review of the literature and the methodological challenges. Neuropsychology Review, 17, 157–177.
Fridriksson, J., Baker, J.M. & Moser, D. (2009). Cortical mapping of naming errors in aphasia. Human Brain Mapping, 30, 2487-2498.
Fridriksson, J., Bonilha, L., Baker, J.M., Moser, D., & Rorden, C. (in press). Activity in preserved left hemisphere regions predicts anomia severity in aphasia. Cerebral Cortex.
Fridriksson, J., Morrow, K. L., Moser, D., & Baylis, G. C. (2006). Age-related variability in cortical activity during language processing. Journal of Speech, Language, and Hearing Research, 49, 690–697.
Fridriksson, J., & Morrow, L. (2005). Cortical activation and language task difficulty in aphasia. Aphasiology, 19, 239–250.
Fridriksson, J., Holland, A.L., Coull, B.M., Plante, E., Trouard, T.P., & Beeson, P. (2002). Aphasia severity: Association with cerebral perfusion and diffusion. Aphasiology, 16, 859-871.
Martin, P.I., Naeser, M.A., Ho, M., Doron, K.W., Kurland, J., Kaplan, J., et al, (2009). Overt naming fMRI pre- and post-TMS: Two nonfluent aphasia patients, with and without improved naming post-TMS. Brain and Language, 111, 20-35.
Meinzer, M., Flaisch, T., Obleser, J., Assadollahi, R., Djundja, D., Barthel, G., et al. (2006). Brain regions essential for improved lexical access in an aged aphasic patient: A case report. BMC Neurology, 17, 6–28.
Postman-Caucheteux, W.A., Birn, R.M., Pursley, R.H., Butman, J.A., Solomon, J.M., Picchioni, D., McArdle, J., & Braun, A.R. (in press). Single-trial fMRI shows contralesional activity linked to overt naming errors in chronic aphasic patients. Journal of Cognitive Neuroscience.
Raboyeau, G., De Boissezon, X., Marie, N., Balduyck, S., Puel, M., Bézy, C., et al. (2008). Right hemisphere activation in recovery from aphasia: Lesion effect or function recruitment? Neurology, 70, 290–298.
Schnur, T. T., Schwartz, M. F., Brecher, A., & Hodgson, C. (2006). Semantic interference during blocked-cyclic naming. Evidence from aphasia. Journal of Memory and Language, 54, 199–227.
Schnur, T.T., Schwartz, M.F., Kimberg, D.Y., Hirshorn, E., Coslett, H.B., & Thompson-Schill, S.L. (2009). Localizing interference during naming: Convergent neuroimaging and neuropsychological evidence for the function of Broca's area. Proceedings of the National Academy of Sciences, 106, 322-327.
Schwartz, M.F., Dell, G.S., Martin, N., Gahl, S., & Sobel, P. (2006). A case-series test of the interactive two-step model of lexical access: Evidence from picture naming. Journal of Memory and Language, 54, 228-64.
Vitali, P., Abutalebi, J., Tettamanti, M., Danna, M., Ansaldo, A.-I.,
Perani, D., et al. (2007). Training-induced brain remapping in chronic aphasia: A pilot study. Neurorehabilitation and Neural Repair, 21, 152–160.
Wilson, S.M., Isenberg, A.L., & Hickok, G. (2009). Neural correlates of word production stages delineated by parametric modulation of psycholinguistic variables. Human Brain Mapping, 30, 3596-3608.