Which Strategies Facilitate Word Retrieval in Non-Fluent Progressive Aphasia? A Single Case Study

Learning Mechanisms Applied to Anomia Treatment in PPA

The literature available to date shows that learning mechanisms used to treat naming deficits in PPA are broad enough to generate an effect that is transversal to all variants of the syndrome [5, 6]. However, the underlying cause of anomia differs across the variants [7] and therefore the learning mechanisms should be tailored to the features of each variant so that a bigger effect size could be accomplished. A scarce number of studies have planned the anomia treatment considering the variant presented by the participants [8, 9, 10]. More often the sample of the studies includes patients diagnosed with different variants of PPA [11, 12, 13], or even different diagnosis, such as progressive aphasia and aphasia resulting from stroke [14, 15]. In this case, the risk might be that the therapy addresses one’s deficits more successfully than the others.

Several learning strategies have been implemented in experimental studies targeting naming skills in PPA. The learning mechanisms can be grouped in the following clusters: i) simple repeated practice; ii) hierarchical cueing method; iii) spaced word retrieval; iv) semantic elaboration of concepts; v) mnemonic generation; vi) errorless vs. errorful learning paradigm; and vii) active vs. passive learning paradigm.

Simple repeated practice was reported in the study of Graham, Patterson, Pratt, & Hodges [16]. A former surgeon diagnosed with svPPA trained forgotten vocabulary through the repetition of lists of words and pictorial encyclopedias that contained labelled pictures, every day, for two weeks. Repeated rehearsal improved name production however, it failed to encourage the access to semantic information related to the trained items. The results of the study suggest that repeated rehearsal might be an option for non-fluent and logopenic variants, but not for semantic variant.

Hierarchical cueing method implements increasing or decreasing cues to prompt picture naming, depending on the study. The cues can be semantic, phonological and/or orthographic-based. An example of this approach is the Lexical Retrieval Cascade Treatment [12], used in two cases, one with the logopenic variant and another with the semantic variant. A sequence of carefully planned tasks engaged and strengthened the patients’ residual language skills, by facilitating the access to the central components of language processing. Increasing cues were provided in the following order: semantic self-cue > orthographic self-cue > phonemic self-cue > oral reading > repetition > semantic plausibility judgements > recall. The two patients showed beneficial effects mirrored by the ability to name the therapy assigned nouns. Following the same line, a model-oriented naming therapy delivered to a case of semantic progressive aphasia failed to reveal significant differences between naming performances on the set with semantic cues, compared to the set with phonological cues [17]. The cueing hierarchy for the phonological set predicted the addition of as many cues as the patient needed to name the target word: number of syllables > first sound > first syllable in the case of multisyllable words or sounds in case of single-syllable words > repetition. The cueing hierarchy for the semantic set comprised four levels: superordinate term > concept definition > semantic closure phrase > listening to the target word.

Spaced word retrieval strategy consists of confrontation naming of the same pictures with increasing time recall intervals, as well as producing the attribute related to each item. Bier and colleagues compared spaced word retrieval with simple word repetition and found no leverage of the first over the second [8]. Both strategies resulted in word relearning for the participant diagnosed with semantic progressive aphasia.

Mnemonic generation was used in the study of Dewar, Patterson, Wilson, & Graham [18], where a subject with svPPA was given photographs of famous people in the United Kingdom and required to name them, one at a time. Each photograph was presented accompanied by the corresponding written word and a semantic fact that were incorporated into a mnemonic. The mnemonic learning mechanism was associated with vanishing cues and expanded rehearsal techniques. After five sessions, the patient was able to recall the target words with a slight reliance on the cues. He also performed much better on category fluency tasks.

Semantic elaboration of concepts is the approach used in the Semantic Feature Analysis (SFA) implemented with a participant that suffered from severe non-fluent progressive aphasia [15]. The participant was required to name pictures representing objects and actions and was prompted with semantic information when he could not find the target word at the first attempt. Subsequent prompts were generated if needed, but after three attempts, the researcher produced the word and asked the patient to repeat. The participant benefited from this treatment, as he correctly named 14 nouns and verbs in a total of 15 after nine sessions. He also showed network expansion and recruited semantic processing areas to name the trained words, as seen in the fMRI performed pre- and post-treatment. The authors suggest that SFA seems to enable brain plasticity changes in degenerative- based anomia, through functional reactivation and reorganization.

A comparison between errorless vs. errorful and active vs. passive learning was conducted by Jokel & Anderson [19]. The participants with semantic variant responded better to errorless learning than errorful. No differences were found between active and passive learning, regarding their role in restoring lexical representations.

Ever since, the evidence pointed to the fact that errorless learning induced more gains than errorful learning, other authors used errorless strategy to treat anomia in PPA [20, 21, 22].

Clinic pathological characteristics of nfvPPA

Neurodegenerative-based aphasia derives from a wide range of pathologies. Among the most typical are frontotemporal lobar degeneration (FTLD), Alzheimer’s disease (AD) and other tauopathies (e.g., corticobasal degeneration, Pick’s disease). The literature shows that up to 30% of nfvPPA cases present AD etiology and approximately 20% of nfvPPA present FTLD etiology. Grossman estimated that the prevalence of nfvPPA due to FTLD is 0.5-3.0 per 100,000 and incidence is 0.4-0.7 per 100,000.

The hallmark of nfvPPA is effortful and halting speech output due to agrammatism along with word-finding pauses [23]. The speech is predominantly marked by grammatical and speech sound errors. While healthy adults typically produce 140 words per minute, these patients generate an average of 45 words per minute [24]. Apraxia of speech is commonly a co-occurring disorder [25]. Impaired syntactic comprehensions is not present at the onset but emerge with the disease progression, as well as, reduced sentence complexity [2]. In these patients, prosody is also affected, as rhythm and melody change. Other clinical features include surface dyslexia and surface dysgraphia. The phonological and syntax dissolution gradually worsens until patients become mute [26] and dependent on assisted communication [27]. Content word comprehension and object knowledge are considered spared.

Researchers have pointed out variable pathology underlying nfvPPA [26]. Progressive nonfluent aphasia might be associated with tau-positive pathology [28]; dementia with Pick bodies pathology [29]; AD pathology [30, 31]; FTLD-U pathology [32]; FTLD-TDP pathology [33]; dementia with Lewy bodies [31]; progressive supranuclear palsy and corticobasal degeneration [34].

When it comes to the PPA management, the most studied variant is the semantic one. There are substantially less studies that investigated the effects of anomia treatment and learning mechanisms in nfPPA. Thus, the aim this study is to determine the learning strategy that offers more gains in patients diagnosed with nfvPPA. Based on the supporting evidence from the literature, we hypothesize that an individual with non- fluent progressive aphasia will respond better to a phonological cueing approach than a semantic one.

Case report

FP is a 60-year-old, right-handed, active Accountant who owes a business with his wife. He is a native speaker of European-Portuguese with a bachelor’s degree.

There was a 2-year history of speech and language deterioration when FP was first diagnosed with nfvPPA associated with apraxia of speech and ideomotor apraxia of the superior limbs, in January of 2017. The Neurologist reported increasing language impairment characterized by non-fluent speech, inconsistent semantic and phonologic errors, and major difficulties in tasks that involve reading and writing. Soft depression symptoms were identified as a comorbidity.

By the time he entered the study, he was completely independent on his daily living. He used to drive by himself, go to work every day, carry out his personal hygiene, shop for groceries at the supermarket, manage the family budget and ride the bike to exercise as suggested by the doctor. As the speech became less fluent and paraphasic, some adjustments were made in his job setting. For example, he kept working at the computer and doing the paperwork but someone else would handle the communication with the clients.

Baseline assessment

A comprehensive evaluation was conducted by a speech-language pathologist (SLP) blind to the experiment, to establish the patient’s neurocognitive and language profile. FP scored 23 (Maximum score=30) at the Mini Mental State Examination (MMSE) [35], showing an initial cognitive impairment. He failed on tasks of orientation to time, word recall, confrontation naming, and sentence writing. A picture description task [36] used to analyze spontaneous speech revealed a non-fluent discourse marked by phonemic, phonologic and semantic paraphasias, word finding pauses, effortful articulation and stereotyped expressions. The application of the Lisbon Aphasia Examination Battery (BAAL) [37, 38] showed impaired object naming and repetition, surface dyslexia and dysgraphia, relatively preserved comprehension of oral commands, and intact object recognition. The patient achieved a 39% percentage of correct answers at the Snodgrass and Vanderwart naming test [39], which showed impaired picture naming. The performance on neurocognitive and language tasks can be found in Table 1.

Procedures

The main purpose of this study was to assess the best strategy for word retrieval in non-fluent progressive aphasia. Therefore, two lists of words were built, one to test naming accuracy after semantic cueing (responsive naming) and the other to assess naming response towards phonologic cueing (first syllable). Each list consisted of 33 items balanced for semantic category. Initially, picture naming was evaluated three times in different days and only the words FP failed to name at least two times in the absence of cues were included in the experimental corpus. The words chosen were objects, food, and animals. Different words took part of each list in order to avoid a repetition effect (Appendix 1). The target items were displayed in the center of a computer screen using a power point presentation. A written cue – semantic or phonological – was placed underneath each colored photograph representing the target words. An example of a semantic cue for the word ‘popcorn’ would be: “I watch a movie and eat [popcorn]”. The phonological cue for the word ‘penguin’ would be “pen-“. FP was required to look at the picture, listen to the cue and then name the target word within 30 seconds. One point was given for each correct answer and zero points were given for each incorrect answer or no response. When the patient produced more than one attempt to name the target word, only the first attempt was considered.

We considered a pre-test, the task in which the patient was required to name a list of words in the absence of cues and post-test, the task in which FP had to name the same list of words with the help of either semantic or phonological cues.

Statistical Analysis of Behavioral Data

Comparison between the pre-test and post-test trials were conducted using the McNemar test (naming score). A p-value of 0.05 was considered. The comparison of naming performance across the three semantic categories was calculated by the mean of naming score achieved in each category at post-test.