Mitigation of Proper Name Retrieval Impairments in Temporal Lobe Epilepsy JoAnn P. Silkes, Ph.D., CCC-SLP Aphasia Research Laboratory Department of Speech and Hearing Sciences University of Washington, Seattle This research was supported by the University of Washington Royalty Research Fund and The Dreuding Foundation.
DISCLOSURE The authors and presenter of this work have no relevant financial or non-financial interests connected with the information presented here.
University of Washington Integrated Brain Imaging Center UW Dept. of Speech and Hearing Sciences Irene Minkina, B.A. Diane Kendall, Ph.D.* Lauren Bislick, M.A. Christina del Toro, Ph.D.** *Also affiliated with the VA Puget Sound Health Care System, and funded by VA R&D Grant #C6572R **Now at Midwestern University (IBIC) UW Dept. of Neurological Surgery, Seattle Children s Hospital Division of Neurological Surgery Jeffrey Ojemann, M.D. UW Depts. Of Radiology and Neurology Thomas Grabowski, M.D. UW-Harborview Regional Epilepsy Center Vaishali Phatak, Ph.D.
Temporal Lobe Epilepsy (TLE) Recurrent, unprovoked seizures Affects 1% of the population May be managed with medications, or If unilateral temporal lobe onset, anterior temporal lobe (ATL) may be surgically removed (Wiebe et al, 2001)
Proper Naming in TLE Seizures generated from left ATL Impaired proper naming(drane et al., 2008; Glosser, Salvucci, & Chiaravalloti, 2003; Griffith et al., 2006; Seidenberg et al., 2002; Viskontas, McAndrews, & Moscovitch, 2002) Other language skills generally intact Recognition (semantic memory) intact
A unique opportunity for treatment? Surgical resection makes problems worse Limited language rehabilitation services typically provided Because impairments can be predicted, could we provide language pre-habilitation? But what would we do?
PDP models of language
Parallel Distributed Processing models of language Language representation is distributed The representation includes Weighted connections within linguistic domains has leaves Semantic feature a Tree in the yard Semantic feature b Weighted connections between linguistic domains tree Orthography Tree In the yard Semantics
Representation of proper names Proper nouns thought to be distributed like common nouns. Actress Oscar Winner Meryl Streep Semantic feature a Semantic feature b Streep Orthography Meryl Streep Actress Semantics So why does focal damage to the left ATL problems with proper name retrieval but not common name retrieval?
ATL and proper name retrieval So why does focal damage to the left ATL problems with proper name retrieval but not common name retrieval? What is special about the retrieval of proper names? Why is recognition of proper names preserved in TLE while retrieval is compromised? What role does the left ATL play in the retrieval of proper names?
Convergence Zone Framework (Damasio, 1989) Left ATL for binding /integration of distributed features for proper names Features distributed across left hemisphere Orthographic, phonologic, semantic, articulatory/motor, etc. Recognition not affected by left ATL seizures Subserved by distributed networks Proper name retrieval depends on binding of linguistic features
How does the convergence zone work? Inherently non-linguistic Binds linguistic features of a given proper name A specific group of hidden units that serves the process of proper name retrieval
Treatment of proper name retrieval: What are our options? Language treatment can be designed to: Strengthen network connections to improve the efficiency of existing network connections or... Shift the system s reliance on a given network to recruit related networks or weaker aspects of the same network. Proper name retrieval treatment left ATL ( binding zone ) is damaged Goal: shift system s reliance from binding zone to distributed features
Treatment of proper name retrieval: Possible mechanisms of change 2 possible outcomes Strengthen distributed features new binding zone outside of the left ATL or Strengthen distributed features proper name retrieval proceeds without binding zone Regardless of mechanism of change, shift of reliance should lead to improved proper name retrieval
Treatment objective Decrease dependence on the compromised region Increase reliance on undamaged networks across left hemisphere Mitigate negative effects of ATL resection on proper name retrieval
Treatment Program Linguistically distributed Lexical Semantic Orthographic Articulatory-motor Phonologic Multi-modality Visual Auditory Articulatory Kinesthetic
Errorless learning Treatment Program Famous faces and famous places One attempt at personally relevant faces
American actress Received awards for her role in Basic Instinct and also Casino Former fashion model
Syllables in her first name: (2) Syllables in her last name (1)
Her First name starts with: /Sh/ Her last name starts with: /S/
Sounds in her first name: 4 Sh ar o n Sounds in her last name: 4 S t o n
Sharon Stone
Who might we want to treat? Individuals who plan to undergo left ATL removal ( prehabilitation ) Individuals who are not scheduled for left ATL surgery (traditional rehabilitation) 3 cases mixed prehab and rehab
Assessment measures for all cases WAB (2/3) BNT Naming probes for trained and untrained items
Treatment Case #1: Pre-habilitation 51 yr old male, epilepsy onset at age 42 Left temporal EEG onset and left mesial temporal sclerosis Previous surgical biopsy for right parietal lymphoma years prior to onset of seizures. BNT 59/60 Impaired proper name retrieval Proper names in social situations Tranel corpus 70% accuracy on famous faces and famous places
Treatment Case #1 Stimuli 32 trained items 10 Famous Faces (FF) 12 Famous Places (FP) 10 Personally Relevant Faces (PRF) 32 untrained control items FF and FP mostly selected from among Tranel s 220 famous faces and famous places that he said were familiar PRF selected from friends, family, and medical team
Treatment Case #1 Baseline testing Treatment Immediate posttreatment testing 32 hours of treatment 2 hours/day 4 days/week 4 weeks Surgery L ATL resection 8 weeks postsurgery testing
Treatment Case #1 Baseline Tx Post-tx 8 weeks post-surgery ES = 26.16 (L) ES = 1.39 (N) ES = 6.17 (L) ES = 1.39 (N) ES = 27.32 (L) ES = 2.8 (S) ES = 6.83 (L) ES = -1.01 (N) ES = 4.36 (M) ES = 1.14 (N) ES =.61 (N) ES = -0.03 (N)
Treatment Case #2: Pre-habilitation 35 yr old male, epilepsy onset at age 33 Left temporal onset, but only subtle anterior and mesial temporal changes on MRI Temporal hypometabolism on PET scan Cortical mapping showed no temporal speech area Post-surgical pathology report showed dysplastic organization of anteriomesial and basal temporal lobe
Treatment Case #2: Pre-habilitation WAB 96/100 BNT 39/60 Tranel corpus - 16.85% accuracy on famous faces and famous places Mild memory impairments on neuropsych testing
Stimuli 10 trained items 5 Famous Faces 5 Famous Places Treatment Case #2 10 untrained control items 5 Famous Faces 5 Famous Places FF and FP selected from among Tranel s 220 famous faces and famous places that he said were familiar
Treatment Case #2 Baseline testing Treatment Immediate posttreatment testing Surgery L ATL resection 6 hours of treatment 2 hours/day x3 nonconsecutive days FF and FP treated simultaneously 11 weeks post-surgery testing
Treatment Case #2 Baseline Tx Post-tx 11 weeks post-surgery ES = 17.32 (L) ES = 4.33 (M) ES = 0 (N) ES = -1 (N)
Treatment Case #3: Rehab AND Pre-hab 61 yr old male, epilepsy onset at age 50 Left temporal EEG onset and left mesial temporal sclerosis h/o anoxia Surgery planned, cancelled, rescheduled 9 months later WAB 97.2/100 BNT 50/60 Impaired proper name retrieval Tranel corpus 70% accuracy on famous faces and famous places
Treatment Case #3 Stimuli 10 trained items 5 Famous Faces 5 Famous Places 10 untrained control items 5 Famous Faces 5 Famous Places All selected from among Tranel s 220 famous faces and famous places that he said were familiar
Treatment Case #3 Baseline testing Treatment Round 2 (3 hrs/2 days) Immediate posttreatment testing Treatment Round 1 (10 hrs/5 days) 9 months posttreatment testing (Baseline for Round 2) Surgery L ATL resection Immediate posttreatment testing 3 months posttreatment testing 6 weeks postsurgery testing
Outcomes Baseline 1 (BL) Treatment Case #3 Tx1 Post-tx 1 (PT) BL2 Tx2 PT2 Postsurgery ES = 3.18 (S) ES = 2.63 (S) ES = 2.02 (N) ES = 2.27 (~ S) ES = 1.5 (N) ES =.75 (N) ES = 1.29 (N) ES = 1.08 (N)
Summary of Findings Participant # of Hours Outcomes #1 32 - Large pre-surgical rehab effects - Med-Large Post-surgical pre-hab effects #2 6 - Large pre-surgical rehab effects, with some generalization - No post-surgical pre-hab effects #3 13 -Small pre-surgical rehab effects - Maintenance at 9 months post-tx -Nearly small post-surgical pre-hab effects
What have we learned from this? This population has potential for improvement in proper name retrieval: Post-treatment improvement : Learning of proper names can occur despite ongoing damage to ATL Post-treatment maintenance after surgery: Learning of proper names can be crystallized into more permanent knowledge
What is the mechanism supporting these changes? Did we reorganize the network for proper name retrieval? Distributed nature of treatment promoted reorganization of proper names to networks outside of left ATL (the binding zone ) How? Focus on distributed linguistic features strengthened undamaged network components Linguistic networks located across left hemisphere were incorporated more strongly into the process of proper name retrieval
Clinical Implications Active treatment of proper name retrieval can be beneficial for patients with temporal lobe epilepsy who: Have no plans to receive surgical intervention Are planning to have ATL resection Have already had ATL resection?? Treatment may be even more clinically meaningful if treated items are individualized and personally relevant Results suggest that high-intensity treatment is needed to instill lasting improvement in function
Replication Predictive factors Dosage Intensity Future directions Populations no surgery, pre-surgical, postsurgical Education, education, education
Thank you for your attention! Questions?
Effect size calculation (Kromrey & Foster-Johnson, 1996) ES = (Mean post-treatment Mean baseline )/SD baseline ]
Control Measures 5-point test cognitive control Low frequency common names linguistic control to show that we re targeting proper name networks rather than the whole language system.
Treatment Cases: Prehab 3 participants (ages: 51, 61, & 35) Treatment dosage P1: 24 hours of treatment in 2-hour segments over four weeks P2: 10 hours of treatment in 2-hour segments over two weeks P3: 6 hours of treatment in 2-hour segments over four days Results All participants showed improved proper name recall immediately after treatment completion (prior to ATL surgery) Post-surgical maintenance P1: Maintenance of trained targets P2: Partial maintenance of trained targets P3: No maintenance of trained targets
Treatment Case #3 Assessment measures Naming probes for trained and untrained famous faces and famous places Naming probes for 145 low-frequency common nouns 5-point test????
Treatment Case #3 Pre-treatment neuropsychological profile Test Subtest Raw Score %ile WMS-III Rey-O Complex Figure Logical Memory Immediate Logical Memory Delay Verbal Pairs Immediate 39/75 63 23/50 75 24/32 84 Verbal Pairs Delay 7/8 84 Faces Immediate 35/48 37 Faces Delay 39/48 84 Copy 33/36 70 Recall 13/36 42
Treatment Cases 1 traditional rehabilitation case surgery 2 pre-habilitation cases