Word retrieval therapies in primary progressive aphasia

Regina Jokel1, Naida L. Graham2, Elizabeth Rochon2, and Carol Leonard3
1Rotman Research Institute, University of Toronto, Toronto, Canada 2Toronto Rehabilitation Institute, University of Toronto, Toronto, Canada 3School of Rehabilitation Sciences, University of Ottawa, Ottawa, Canada
Background: Primary progressive aphasia (PPA) with its three variants is a progressive neurodegenerative dementia in which language impairment is the first and most domi- nant symptom. Traditionally, speech-language pathologists who deliver therapy to adults with acquired neurogenic language disorders shy away from treatment of pro- gressive aphasia as there is no promise of lasting effects and only limited data regarding treatment efficacy. Aims: This paper comprises the most current review of the literature focused on treat- ment of naming impairments in PPA, and aims to encourage and assist clinicians in selecting intervention approaches for individuals with PPA. It highlights current trends and challenges in delivering successful therapy for naming deficits in PPA. Main Contribution: We reviewed papers that reported different forms of naming therapy for patients with PPA, which included interventions that, although not always aimed directly at anomic deficits, brought about improvement in naming. Immediate gains, maintenance, and generalisation effects are summarised, along with a variety of approaches and methodologies that can be applied to the PPA population. We also provide a list of factors that were found to contribute to the success of therapy and to the maintenance and/or generalisation of treatment gains. Conclusions: Current literature delivers encouraging evidence for clinicians wanting to provide naming therapy to patients with PPA. Although PPA is a progressive disorder, both the immediate treatment effects and, in many cases, maintenance results show that improvements are possible. The issues of generalisation of naming gains beyond the clinician’s office still require more studies to determine the best conditions, designs, and patient suitability.
Keywords: Logopenic progressive aphasia; Semantic progressive aphasia; Non-fluent/ agrammatic progressive aphasia; Anomia therapy; Maintenance; Generalisation.
Language abilities profoundly affect our capacity to participate in most everyday activities and engage in meaningful relationships. Consequently, a language impair- ment has a significant adverse effect on a person’s quality of life. Language impair- ment is a core feature of the disorder known as primary progressive aphasia (PPA). Currently, three distinct variants of PPA are recognised: (a) semantic (svPPA), (b) non-fluent/agrammatic (nfvPPA), and (c) logopenic (lvPPA) (Gorno-Tempini et al.,
Address correspondence to: Regina Jokel, Rotman Research Institute, University of Toronto, 3560 Bathurst Street, Toronto, ON, Canada M6A 2E1. E-mail: [email protected]
Aphasiology, 2014 Vol. 28, Nos. 8–9, 1038–1068, http://dx.doi.org/10.1080/02687038.2014.899306
© 2014 Taylor & Francis
2011). These variants are differentiated by the fluency, grammatical correctness, and rate of connected speech, as well as by the status of the semantic (i.e., word meaning) and phonological (i.e., word form and sound) stores.
The semantic variant of PPA (svPPA), also known as fluent progressive aphasia or semantic dementia, is associated with difficulty in understanding word meaning and/ or object identity (Hodges, 2001; Hodges, Graham, & Patterson, 1995; Hodges, Patterson, Oxbury, & Funnell, 1992). As a result of semantic loss, the accompanying language disorder is characterised by progressively empty, though fluent and well- articulated speech, and whole word substitutions called semantic paraphasias (e.g., potato→“apple”) (Breedin & Saffran, 1999; Hodges, Patterson, & Tyler, 1994; Neary et al., 1998; Warrington, 1975).
The non-fluent/agrammatic variant of PPA (nfPPA) is characterised by dissolution of language form (i.e., phonology) and/or syntax (i.e., grammar), in the absence of semantic deficits (Croot, Patterson, & Hodges, 1998, 1999; Mesulam & Weintraub, 1992; Watt, Jokel, & Behrmann, 1997). Errors in the spontaneous speech of people with nfvPPA consist predominantly of sound substitutions called phonemic para- phasias (e.g., apricot→“hipracot”), and they may be accompanied by agrammatism and/or apraxia of speech.
The logopenic variant of PPA (lvPPA) is a relatively new diagnostic entity char- acterised by anomia, simplified grammar, poor oral repetition of sentences, and a phonological impairment (Gorno-Tempini et al., 2004, 2011). Similar to nfvPPA, the errors in spontaneous speech are predominantly phonological in nature. Logopenic PPA has not been well studied, and at present, limited evidence of successful treat- ment for anomia is available for individuals with this variant.
Within the language rehabilitation literature, there is an abundance of studies on therapy after stroke, but reports on structured impairment-based approaches to treating language deficits in individuals with PPA are few in number. And yet, there is evidence to suggest that individuals with mild PPA may be very good candidates for language rehabilitation. In particular, hippocampal integrity has been well documented in early PPA (Gorno-Tempini et al., 2011) and this is con- sistent with the clinical and experimental observation that episodic and autobiogra- phical memory is initially preserved (Snowden, Griffiths, & Neary, 1994, 1999). Thus, individuals with mild PPA should still possess the necessary cognitive prere- quisites to engage in active therapy, and as will be demonstrated in the following overview, the potential for successful retraining is promising.
Two excellent recent reviews of therapy addressing various aspects of language dysfunction in PPA have recently become available (Carthery-Goulant et al., 2013; Croot, Nickels, Laurence, & Manning, 2009). However, our review is focused specifically on therapies which target naming impairment (anomia), the most perva- sive initial symptom of PPA. It should be noted that while a small number of papers reported improvements in verbal expression that included improved naming (MC in Cress & King, 1999; DD in Murray, 1998; RP in Thompson & Johnson, 2006), the treatment approach itself did not target anomia and is therefore not included in the following overview.
This overview zeroes in on factors that were deemed relevant to clinical practice focusing on therapy for anomia in PPA, and includes participant characteristics, the designs and approaches used, the immediate and delayed effects of treatment, gen- eralisation effects, the use of pharmacology and technology in treatment, and the results of neuroimaging studies investigating the influence of anomia treatment on
THERAPY FOR WORD FINDING IN PPA 1039
neural functioning. We selected 39 publications, both full papers and peer-reviewed abstracts, targeting anomia intervention in PPA. Electronic databases, PsycINFO, Medline, Embase, and PubMed, were searched from 1982 (date of the very first publication on PPA by Mesulam) to 2013, using combinations of the following search terms: therapy for anomia, naming therapy, progressive aphasia, logopenic aphasia, non-fluent progressive aphasia, semantic dementia, fluent progressive aphasia, maintenance, generalisation, and treatment gains. In the field of rehabilitation, this may not be a large number. However, compared to only a single report available several years ago (McNeil, Small, Masterson, & Fossett, 1995), rehabilitating anomia in PPA has made significant strides.
PARTICIPANT CHARACTERISTICS
We attempted to identify characteristics that would define the best candidates for treatment. Participants in the reviewed papers included both men and women from different ethnic backgrounds, with a wide age bracket ranging from 30 (Kertesz et al., 2008) to 80 years (Boxer et al., 2013; Kertesz et al., 2008) and a wide range in level of education (ranging from limited formal education to a doctoral degree). Time post- onset usually approximated two years as this is a critical period when patients with PPA do not exhibit other cognitive impairments and can be reliably diagnosed (Mesulam, 1982). However, some studies reported no cognitive problems other than language as late as 6 years post-onset (e.g., Jokel & Anderson, 2012; Jokel, Rochon, & Leonard, 2006). The severity of naming impairments was reported in some cases and ranged from mild (e.g., Jokel, Cupit, Rochon, & Leonard, 2006) to severe (e.g., Cress & King, 1999), sometimes with varied levels of severity within the same study (Savage, Ballard, Piguet, & Hodges, 2013). The majority of studies focused on semantic dementia (e.g., Dressel et al., 2010; Green Heredia, Sage, Lambon Ralph, & Berthier, 2009); however, there were also a few that focused on nfvVPPA (e.g., Jokel, Cupit, Rochon, & Leonard, 2009), and two that focused on the newest—logopenic variant (Beeson et al., 2011; Trebbastoni, Raccah, deLena, Zangen, & Inghilleri, 2013). As lvPPA has been recognised only since its introduction by Gorno-Tempini and her colleagues (Gorno-Tempini et al., 2004), it is possible that some of the treatment studies preceding the year 2004 may have been applied to patients with lvPPA under the diagnosis of nfvVPPA or non-specified PPA. It should be noted that in the body of the paper, we refer to the three PPA variants according to the new nomenclature; however, in the Table (Appendix) we left the original classification used by the authors of included papers to preserve the original way in which the participants were characterised. We attempted to examine which partici- pant characteristics appeared to coincide with best treatment outcomes, but due to missing information (e.g., time post-onset) and differences in approaches and mea- sures, this analysis did not yield useful or reliable information.
DESIGNS
Perhaps not surprisingly, given the relatively low prevalence of PPA, treatment studies with this population usually involved a small number of participants. In fact, most studies included one (e.g., Beeson et al., 2011; Jokel, Rochon et al., 2006) to three participants (e.g., Henry, Beeson, & Rapcsak, 2008, 2009; Senaha, Brucki, & Nitrini, 2010). The methodology prevalent in these studies—and certainly
1040 JOKEL ET AL.
appropriate for a treatment study—employed a pre–post-treatment design, with some including the additional control afforded by a single-subject design. For instance, Cartwright and Elliot (2009) employed an alternating treatment design, Dressel et al. (2010) an ABBA design, Finocchiaro et al. (2006) an ABA design, and Bier an ABCBCBCA design, and several studies have employed a multiple baseline across behaviours design (e.g., Henry et al., 2008, 2009; Jokel & Anderson, 2010; Jokel, Rochon, & Anderson, 2010; Schneider, Thompson, & Luring, 1996).
Thus far, five published studies employed group designs involving participants with PPA, four of which addressed pharmacological intervention with medications that had been previously successful as cognitive enhancers in dementia of the Alzheimers type. One investigated the efficacy of Bromocriptine in six participants with PPA (variant not specified), employing a double-blind, placebo-controlled cross- over design (Reed, Johnson, Thompson, Weintraub, & Mesulam, 2004). Another Bromocriptine study included patients with all three variants of PPA (Farrajota et al., 2012). In this study, 10 patients in each of a treatment and a control group were enrolled in an open-label parallel prospective longitudinal study across two sites. A randomised, placebo-controlled 8-week treatment with Galantamine was conducted with 36 participants with (non-specified) PPA (Kertesz et al., 2008). A similar number of participants with frontotemporal dementia (FTD)/svPPA (37) were enrolled in a randomised, double-blind, placebo-controlled 26-week trial of Memantine (Boxer et al., 2013). The only group study that addressed naming via behavioural treatment was published by Jokel and Anderson (2012), in which four different therapy conditions were tested in seven patients with svPPA. This study employed a one-group pre-test–post-test design.
While pharmacological studies may have an advantage over single-case studies with respect to the number of participants (i.e., these are usually large group studies), they also have the disadvantage that the variant of PPA is often not specified and there is a lack of clinically relevant characterisation of patients’ language and cognitive profiles. Consequently, these studies offer little guidance to clinicians when attempting to fine tune their treatment approach to a PPA variant.
BEHAVIOURAL TREATMENT APPROACHES
Generally, treatment studies for anomia used semantic (e.g., description of perceptual features, purpose, location) and/or phonological (e.g., first or last sound, rhyming words) approaches. Semantically based approaches were reported more frequently due to the fact that patients with semantic dementia are more common as study participants than those with nfvPPA (but see Green Heredia et al., 2009 and Jokel, Cupit, et al., 2006 for exceptions). Three studies explicitly combined or compared the two treatment approaches (Bier et al., 2009; Dressel et al., 2010; Jokel & Anderson, 2010), and several also incorporated reading (Green Heredia et al., 2009; Jefferies, Bott, Ehsan, & Lambon Ralph, 2011; Jokel & Anderson, 2008; Jokel et al., 2010; Mayberry, Sage, Ehsan, & Lambon Ralph, 2011; Newhart et al., 2009; Snowden & Neary, 2002), both with (Green Heredia et al., 2009; Jokel, Cupit, et al., 2006) and without (Mayberry et al., 2011; Newhart et al., 2009; Snowden & Neary, 2002) the use of a computer program. Bier et al. (2009) utilised spaced retrieval as a successful treatment option in PPA. Spaced retrieval was initially developed for patients with Alzheimer’s disease to help them remember new information by increasing the intervals of time between recall of information being learned. Although mostly not
THERAPY FOR WORD FINDING IN PPA 1041
specified (but see Dressel et al., 2010), a few of the studies appeared to use an intense treatment protocol and others appeared to use an errorless approach (e.g., Frattali, 2004; Jokel & Anderson, 2012). Errorless learning is an approach to treatment whereby better outcomes are hypothesised to occur because of prevention of errors in the learning process. Lastly, one longitudinal study reported using a multi-mod- ality stimulation approach with PPA (Farrajota et al., 2012).
One of the questions arising from successful treatment outcomes in svPPA is whether improvements are due to relearning of the word’s phonological forms or re-instatement of semantic information. Jokel and Anderson (2010) utilised both semantic and phonological cues in two participants with svPPA (MK and CW) in a crossover design. They found that while for MK semantic cues were much more successful in improving picture naming, for CW both types of cues were equally successful. Because semantic cues in their approach also contained phonological information (i.e., object names in addition to their semantic descriptors), it could be concluded that, perhaps, CW could only benefit from phonological information and restored picture names only while MK was more successful in rebuilding both names and semantic representations for his words. Assessment of generalisation of treatment gains would confirm this conclusion.
Initially preserved cognitive skills allow individuals with PPA to take part in computer-based interventions. Moss Talk Words® software (Moss Rehab, Philadelphia, PA, USA; Fink, Brecher, Montgomery, & Schwartz, 2001) was used to treat anomia in two individuals with nfvPPA (Jokel et al., 2009) and in one with svPPA (Jokel et al., 2010). This program was originally developed for naming therapy in post- stroke aphasia and is comprised of 340 words with corresponding pictures and cues (both spoken and written). Results indicated that all three participants in the two studies showed significant improvement on naming accuracy following treatment, as well as modest maintenance effects without further practice. This suggests that soft- ware-based therapies may be successfully adapted for use with individuals with PPA.
EFFECTS OF BEHAVIOURAL TREATMENTS OF ANOMIA IN PPA
Behavioural treatment of anomia in PPA was shown to produce immediate gains in all of the studies reviewed. Those gains were brought about, most often, by a phonological approach in nfvPPA patients (e.g., Jokel et al., 2009), and a semantic approach in those with svPPA (e.g., Jokel & Anderson, 2012). This preference for one of the two approaches was dependent on patients’ core impairment and the notion adopted from the post-stroke literature that impairments in phonology (such as in nfvPPA) will benefit from a phonological approach while semantically based disorders (such as svPPA) will respond better to semantic treatment. However, in Jokel, Rochon, et al. (2006), Jokel and Anderson (2008, 2012), all participants with svPPA showed significant improvement on naming accuracy in both the phonologi- cal and semantic conditions raising the notion that in addition to semantics, other factors may influence success in svPPA. In particular, personal relevance and invol- vement of the client in the selection of stimuli may have been factors contributing to the success of language therapy in these and other studies (e.g., Bier et al., 2009; Snowden et al., 1994; Snowden & Neary, 2002). Partially spared semantic knowledge (e.g., Jokel, Rochon, & Leonard, 2002; Snowden & Neary, 2002) and re-training of daily vocabulary by incorporating it in daily conversations also proved helpful (e.g., Green Heredia et al., 2009). All of these findings point to the benefits of utilising
1042 JOKEL ET AL.
meaningful and/or functional stimuli in treatment. In view of the progressive nature of PPA, and the urgency to maintain the existing vocabulary, perhaps only those items that are relevant to daily life should be included in treatment sets.
MAINTENANCE
Approximately one third of the studies reviewed (12 out of 39) did not evaluate or report the maintenance of treatment gains beyond the immediate post-treatment period. Of those that did assess maintenance, the time interval between end of treatment and follow-up testing varied substantially, generally from 1 week to 6 months. The results of maintenance effects varied. Many studies report some degree of maintenance (e.g., Beeson et al., 2011; Bier et al., 2009; Green Heredia et al., 2009; Jokel & Anderson, 2009; Dressel et al., 2010; Savage et al., 2013; Snowden & Neary, 2002) while others do not (Frattali, 2004; Louis et al., 2001; Trebbastoni et al., 2013). Although the results of maintenance effects were mixed, overall, when maintenance was measured, there were more studies that documented maintenance than not (see Appendix).
Some studies (i.e., Jokel, Rochon, et al., 2006; Jokel et al., 2010) showed that practising items that are still in the patient’s vocabulary may be beneficial to main- taining those items. Similarly, partially spared knowledge of items’ semantics was found to facilitate both immediate treatment gains and maintenance effects (Snowden & Neary, 2002). The fact that maintenance effects have been found is impressive, given the progressive nature of the disorder (as noted by McNeil et al., 1995). Inspection of Appendix shows that the effects of treatment have been maintained, when reported, from 1 week to 6 months later; although 1 month seems to have been the most frequent time point for follow-up testing. Interestingly, the maintenance results do not appear to be dependent upon the PPA variant; however, it must be remembered that there have been many more treatment studies involving patients with svPPA than with either nfvPPA or lvPPA. In addition, in the context of a progressive disorder it may be important to note that two types of improvement should probably be evaluated—namely the difference between baseline and post- treatment gains and the rate of forgetting of trained items compared to the untrained ones.
GENERALISATION/FUNCTIONAL GAINS
Generalisation following language therapy is defined as the use of trained items or strategies in untrained communicative contexts. The ultimate goal of therapy is improvement which will generalise to functional communication in everyday life, but this is difficult to measure and evaluation typically relies upon self-reports from the patient and/or their family. Generalisation can, however, be systematically mea- sured in a number of ways in a therapy or research situation. The types of general- isation which have been evaluated for potential improvement following therapy for anomia in PPA are the following: (1) naming of treated items in contexts which are minimally different from the therapy context, (e.g., the items are presented in a different order, or on different coloured paper); (2) naming of alternative exemplars of treated items (visually similar, or visually dissimilar, to the treated exemplars); (3) naming of untreated items, often with a focus on items from the same semantic category as the treated items; (4) use of treated items in untreated contexts, such as
THERAPY FOR WORD FINDING IN PPA 1043
category and/or letter fluency, sentence production tasks, or narrative speech tasks; and, less frequently, (5) performance in other domains is examined for change following therapy for anomia (e.g., syntactic production, or performance on an aphasia examination). As outlined below, the results with respect to generalisation are different for each variant of PPA.
Semantic variant PPA
In general, relearning in svPPA is strongly context dependent and does not generalise to untrained items or tasks (Bier et al., 2009; Dressel et al., 2010; Frattali, 2004; Graham, Patterson, Pratt, & Hodges, 1999; Graham, Patterson, Pratt, & Hodges, 2001; Jokel, Cupit, et al., 2006; Jokel et…