Lexical retrieval treatment in primary progressive aphasia: An investigation of treatment duration in a heterogeneous case series

Abstract

Word-finding difficulty is typically an early and frustrating symptom of primary progressive aphasia (PPA), prompting investigations of lexical retrieval treatment in PPA. This study aimed to investigate immediate treatment gains following two versus four weeks of treatment, item generalisation, and maintenance of gains with ongoing treatment in a single case series of eight individuals with heterogeneous PPA presentations (three non-fluent/agrammatic, two logopenic, two semantic, and one mixed PPA). Three individuals made initial gains in picture naming and maintained them over 6 months or more with ongoing treatment. By contrast, three individuals made marginal initial gains but were unable to continue treatment, and two individuals did not make the typically-reported initial gains with two or four weeks of treatment. There was little evidence of generalisation to untreated items. Our results add to the evidence that daily home practice of Repetition and Reading in the Presence of a Picture over extended periods can increase and maintain retrieval of personally-relevant words in picture naming for some individuals with semantic or nonfluent/agrammatic variant PPA. Further research is needed into the factors associated with long-term treatment adherence and gains, and the factors associated with nonadherence to treatment.

Introduction

The primary progressive aphasias (PPAs) are clinical syndromes associated with frontotemporal lobar degeneration or Alzheimer's disease neuropathology. They are characterised by prominent language impairments with relative sparing of other cognitive abilities in the early years post-onset. The three clinical variants recognised in a consensus paper (Gorno-Tempini et al., 2011) are nonfluent/agrammatic-, semantic¹- and logopenic-variant PPA, and there are further “mixed” cases that do not meet diagnostic criteria for these variants (Sajjadi, Patterson, Arnold, Watson, & Nestor, 2012). Difficulty in word retrieval is typically an early and frustrating symptom (Mesulam, 2001; Rohrer, Knight, J. E Warren, Fox, Rossor & J. D Warren, 2008), leading individuals with PPA to seek diagnosis and treatment (Rising, 2014), and prompting increasing research on lexical retrieval treatment in PPA over the last decade (Croot, 2018b).

A systematic review of nonpharmacological interventions for cognitive impairment in PPA (Carthery-Goulart et al., 2013) found that lexical retrieval treatment is by far the most common intervention conducted with people with PPA. Evidence of lexical retrieval treatment efficacy is needed particularly in the nonfluent/agrammatic and logopenic PPA variants (Carthery-Goulart et al., 2013), and more research is needed on the treatment and participant factors supporting optimal immediate gains, maintenance, and generalisation of gains in all three PPA variants, to inform clinical decisions about which treatments are best suited to which individuals (Cadório et al., 2017, Carthery-Goulart et al., 2013, Croot, 2018b, Croot et al., 2009, Henry et al., 2008, Jokel et al., 2014).

The present study aimed to add to the evidence for immediate gains, maintenance of gains with long term treatment, and generalisation of gains to untreated items in lexical retrieval treatment, in an experimental single case series of eight individuals with heterogeneous clinical presentations of PPA. The treatment technique was Repetition and Reading in the Presence of a Picture, and participants were an unselected convenience sample of individuals drawn from working PPA clinics in Sydney, Australia, and Munich, Germany. The heterogeneous nature of the sample allowed us to observe a range of treatment outcomes and adherence patterns under the same treatment protocol, and to describe disease and participant factors associated with these outcomes. This Introduction provides an overview of research on, and gaps in our understanding of lexical retrieval treatment outcomes in PPA, and the rationale for and aims of the study.

We define “immediate treatment gains” in this paper as changes in the behaviour targeted by the treatment that are attributed to the treatment (see also Croot et al., 2009); these have been described elsewhere as “direct treatment gains” (Beales et al., 2016, Henry et al., 2013). In lexical retrieval treatment, such gains are demonstrated by improvement in retrieval of treated items, typically in picture naming, sometimes in word generation (category fluency) tasks. Immediate treatment gains are reported in almost all published studies of lexical retrieval in PPA (Cadório et al., 2017, Carthery-Goulart et al., 2013, Croot et al., 2009, Jokel et al., 2014).

Henry et al. (2008) suggested that immediate treatment gains are a function of disease factors such as PPA subtype and severity, the associated distribution of pathology, and the residual memory and learning mechanisms. Treatment factors such as the items, language tasks and learning mechanisms targeted in treatment (Beales et al., 2018, Jokel and Anderson, 2012, Wilson, 2002), the therapy materials and activities, and the amount of practice are also likely to contribute to outcomes (Croot et al., 2009, Jokel et al., 2014). Savage, Ballard, Piguet, and Hodges (2013) found that level of treatment gain was similar over three versus six weeks of treatment pairing a picture with its name and a personally-meaningful audio description for two people with semantic variant PPA, raising the intriguing possibility that increasing treatment duration may not increase treatment gain. Jokel et al., 2014 critical review concluded that success was likely to be greater with personally familiar items, and that in semantic variant PPA preserved knowledge of treated items was associated with greater improvement. Nevertheless, it is not yet clear which combination of disease, participant and treatment factors are associated with optimal immediate treatment gains.

There are similar questions about which factors are associated with the maintenance of immediate treatment gains over time, with and without ongoing treatment.

Our 2009 review of published behavioural interventions in PPA (Croot et al., 2009) found consistent loss of immediate treatment gains without ongoing treatment over two to twelve months following cessation of treatment activities, in the 14 studies that had investigated maintenance at that time. The one exception was an individual with semantic variant PPA whose husband reported both generalisation of treated-word use to everyday conversation, and 8-month maintenance of gains without ongoing structured practice (Heredia, Sage, Ralph, & Berthier, 2009). Her maintenance of gains was tentatively attributed by the researchers to the inclusion of personally meaningful definitions in treatment, her rapid learning (allowing a long period of consolidation after acquisition during the treatment period), and her ongoing use of the treated words in everyday situations after treatment. Our review recommended continuing treatment activities to maintain lexical retrieval for as long as treatment is appropriate in the context of other aspects of the individual's care, treatment goals, and life goals (Croot et al., 2009), a position also endorsed by other researchers (Cadório et al., 2017, Jokel et al., 2014).

In the five years following 2009, the number of lexical retrieval studies investigating maintenance of treatment gains almost tripled. Jokel et al. (2014) found that approximately two thirds of the 39 studies they reviewed had investigated maintenance of treatment gains, and more than half of these reported maintenance without ongoing treatment, although some only reported maintenance over periods of a week. At present, differences in study design and participant characteristics make it difficult to determine which participant and treatment factors are reliably associated with maintenance of improvements without ongoing treatment (Carthery-Goulart et al., 2013). In the study comparing three versus six weeks' treatment, mentioned earlier, Savage, Ballard, et al. (2013) found better maintenance of treatment gains after six than three weeks' practice for two people with semantic variant PPA.

Very few studies have been conducted to investigate maintenance of lexical retrieval with ongoing treatment over periods of 6–24 months. Savage, Piguet, and Hodges (2015) reported nine participants with semantic dementia whose maintenance of word retrieval in picture naming was monitored over six months following two initial four-week periods of treatment. When naming accuracy dropped below 80% of post-treatment levels, revision practice was introduced to restore post-treatment accuracy. There were three broad outcomes: participants maintained accuracy at or above 80 per cent of words gained over the treatment period with no revision required (three participants); they required ten or fewer revision sessions over the four-month monitoring period to maintain the accuracy of words gained in treatment at or above 80% (four participants); or they required continuous revision after 2 months (two participants). Maintenance with less revision practice was associated with less severe disease, and all participants were able to maintain 80 percent or more of their treatment gains over six months with more or less revision.

Rogalski et al. (2016) offered lexical retrieval treatment (semantic, phonological and orthographic cueing) alone or in combination with motor speech treatment via an online/telemedicine platform to 21 individuals with varied PPA presentations. Around half the participants continued the treatment for 6 months, with production accuracy for this group maintained over this period.

Meyer, Tippett, and Friedman (2016) and Meyer, Tippett, Turner, and Friedman (2018) reported on six months of orthographic/phonological treatment for words that were correctly named over three baselines (prophylaxis items) and words incorrectly named over three baselines (remediation items) for groups of people with semantic variant, logopenic variant and, for prophylaxis items only, nonfluent/agrammatic variant PPA.² In the orthographic treatment, participants saw the written name and picture separately and in combination for each item, and were asked to produce the name in written and spoken form and to recognise the stimulus. In the phonological treatment, the written stimuli and task were omitted. One month after the end of the treatment period they found less decline on treated prophylaxis items, consistent with the suggestion by Jokel et al. (2014) that ongoing practice would delay decline of known items. In several treatment conditions and naming tasks, Meyer and colleagues also found more improvement on treated remediation items than on untreated items, for semantic and logopenic PPA. Only one individual with nonfluent/agrammatic variant PPA had remediation items, and no statistical analysis is reported for these items, but visual analysis suggests there was also improvement in written naming and perhaps in one or more spoken naming condition one month after the end of six month's treatment for this individual. Meyer et al. (2018) also investigated maintenance of gains on remediation items without ongoing treatment in the same study, a further 8 and 15 months later. On average the semantic and logopenic group remained more accurate than at pretreatment in one of two spoken confrontation naming conditions, but only for items which had received orthographic treatment. Otherwise, spoken confrontation naming, written naming and naming in a scene description task (in which the treated item was shown in a photograph of a scene) were not significantly different on average across this group than at pretreatment. It is unclear whether the gains in spoken naming in the orthographic treatment condition for the individual with nonfluent/agrammatic PPA were reliably maintained above baseline in the absence of a statistical analysis.

Finally, Reilly (2016) reported the longest treatment to date for five people with semantic variant PPA, using pictures organised in semantic categories on a communication board. In 30-min sessions approximately three times a week, a clinician gave the name of a picture, asked the participant to repeat it, then asked them to name the picture with no further cue, initiating a series of naming attempts of correctly-named items at increasing intervals to elicit spaced retrieval. The participant who was most impaired on entry to the study withdrew after 8 months with no treatment gain, but three participants practised for 18–19 months and one for 24 months. Although there was only a single baseline assessment in this study and there is no statistical analysis of treatment-specific effects, naming accuracy approximately doubled for two individuals and increased by approximately 25% for the two others on 100 items, after ongoing treatment for eight (one participant) or 16 months (three participants). In general, participants showed rapid gains, a plateau, then gradual decline.

In summary, studies investigating maintenance of treatment gains suggest that retrieval accuracy can be maintained (prophylaxis items) or improved (remediation items) with long-term (six months or more) continuous treatment (Meyer et al., 2018, Meyer et al., 2016, Meyer et al., 2016, Reilly, 2016, Rogalski et al., 2016, Savage et al., 2013a). These studies also offer two caveats: first, some individuals may not persist with treatment activities over the 6-month period (Reilly, 2016, Rogalski et al., 2016), and second, over a sufficiently long period, decline may become evident even with ongoing treatment (Reilly, 2016). These studies indicate that a longer duration of treatment may be associated with better maintenance of treatment gains once treatment ceases in semantic variant PPA (Savage, Ballard et al., 2013), that gains may be maintained above baseline levels over at least 16 months with ongoing treatment in semantic variant PPA (Reilly, 2016), and that items that can be retrieved at the start of treatment (prophylaxis items) will be maintained longer with treatment (Meyer et al., 2018). There is minimal information on long-term maintenance of gains on items that could not be named reliably before treatment in nonfluent/agrammatic variant PPA, nor on whether the group effects reported by Rogalski et al. (2016) and Meyer et al. (2018) are reliable for all treated individuals in those groups.

Studies of lexical retrieval treatment in PPA also address generalisation of treatment gains to untreated items and untreated language tasks such as sentence production, video or scene description or semi-structured interview. The current study investigated item generalisation, which has previously been investigated in 34 reports of lexical retrieval treatment in PPA, with 48 of the 66 individuals in these reports showing no item generalisation (Croot, 2018a). Item generalisation is also rare in the literature on lexical retrieval treatment following stroke (Nickels, 2002a).

Item generalisation for 18 individuals was reported in nine studies, but the statistical analysis was not included for individual participants in three of these (Jokel and Anderson, 2012, McNeil et al., 1995, Meyer et al., 2016), and the results were statistically unreliable in three others (Newhart et al., 2009, Robinson et al., 2009, Savage et al., 2014), leaving only three of 34 reports with potentially persuasive generalisation (Beales et al., 2016, Beeson et al., 2011, Henry et al., 2013). Our earlier study of two individuals with PPA undergoing the same treatment we report here found no generalisation to untreated words (Croot et al., 2015). Cadório and colleagues in their (2017) systematic review of generalisation of PPA treatment outcomes in 25 studies recommended that future studies investigate item generalisation to determine whether there is reliable evidence for generalisation.

Carthery-Goulart et al. (2013) rated the quality of evidence in reports of PPA treatment in their systematic review, using criteria described by Cicerone et al. (2000). The vast majority of studies were rated as Class III, “obtained from one or more single cases that used appropriate single-subject methods with adequate quantification and analysis of results” (Carthery-Goulart et al., 2013, p. 127). They concluded that together, the reviewed studies provided sufficient evidence to support lexical retrieval treatment as a “Practice Option”³ in semantic variant PPA. There was insufficient investigation of lexical retrieval in nonfluent/agrammatic PPA and logopenic PPA to fully support such a recommendation, despite positive reports of treatment effects. Therefore, despite the burgeoning number of studies on lexical retrieval treatment in PPA, evidence on lexical retrieval treatment in nonfluent/agrammatic PPA and logopenic PPA is still limited (Carthery-Goulart et al., 2013).

Given the heterogeneity of disease, participant and treatment factors across studies, it also remains difficult to establish the combination(s) of factors associated with optimal immediate gains, maintenance and generalisation of gains. One approach is therefore to restrict the treatment factors and study the effects of a single treatment across a case series of diverse individuals. This approach is taken in this study, as well as in previous reports each describing a series of participants (e.g., Jokel et al., 2006, Reilly, 2016, Savage et al., 2015).

It remains an open question in the non-progressive aphasia literature as to whether providing the treatment target in written or spoken form for the individual with aphasia to reproduce is as effective as requiring retrieval from lexical memory (Middleton & Schwartz, 2012), and we chose the former approach for a number of reasons. First, lexical retrieval in picture naming activates a wide network of language regions in the brain (Levelt, Praamstra, Meyer, Helenius & Salmelin, 1998). This activation may be amplified in treatments such as Repetition and Reading in the Presence of a Picture that provide multimodal information about the target lexical items (semantic, in the picture, phonological, in the spoken model to be repeated, and orthographic, in the written label), and may support function of the picture naming or wider language network in the progressive aphasias, where there is reduced activation in regions of the language network distal to the primary sites of atrophy (Mummery et al., 1999). Prophylactic effects of lexical retrieval treatment (Jokel et al., 2014, Meyer et al., 2018, Meyer et al., 2016) may arise because promoting activation of those regions supports their function over a longer period. Second, we have previously reported immediate treatment gains using this treatment with two people with PPA (one nonfluent/agrammatic PPA, one logopenic PPA; Croot et al., 2015), and it has been used successfully with semantic variant PPA (Savage et al., 2013a, Savage et al., 2014, where it was described as “Look, Listen, Repeat” treatment). Savage, Ballard, et al. (2013) compared Look, Listen, Repeat treatment with the same treatment incorporating a sentence generation task in one study and a semantic description in another, and found no advantage in lexical retrieval outcomes for either of the latter two conditions, suggesting that a range of potential cognitive mechanisms involved in sentence generation (including unprompted retrieval) or semantic elaboration did not yield additional treatment gains for the reported individuals with semantic variant PPA. Third, Repetition and Reading in the Presence of a Picture has the advantage of being relatively simple for a person with PPA to carry out as a home treatment, increasing the person's likely adherence to treatment, especially over the long term. Home treatment also allows for higher treatment intensity, as treatment can be undertaken every day without daily visits to the clinic. Daily treatment in turn provides distributed practice, which is likely to improve retention of learning (Middleton & Schwatz, 2012; Raymer et al., 2008). Finally, Repetition and Reading in the Presence of a Picture is relatively simple for the clinician to prepare, increasing the likelihood of translation of any positive results of our study to clinical practice.

Our first aim was to detect immediate treatment effects on picture naming of treated items in a heterogeneous sample of people with PPA as a result of home practice using Repetition and Reading in the Presence of a Picture. Our second was to compare the treatment gains evident after two weeks versus four weeks of treatment to investigate whether longer treatment resulted in greater immediate treatment gains. Savage, Ballard et al. (2013) found no difference in immediate treatment gains obtained over three-versus six-week treatment periods, suggesting that gains had plateaued within three weeks of starting treatment. We therefore manipulated treatment duration over a shorter period than Savage, Ballard et al. (2013). Finally, we aimed to deliver a period of ongoing treatment over at least six months. We sought to investigate whether continuous treatment over the long term would maintain word retrieval gains the participants achieved in the earlier treatment periods, providing evidence for our and others' clinical recommendation in support of ongoing treatment to maintain retrieval of communicatively important vocabulary in PPA.

We hypothesised that, as typically reported, we would see immediate treatment gains for all participants, regardless of PPA variant and clinical presentation. We also hypothesised that we would see greater immediate treatment gains after four than after two weeks' treatment. Our third hypothesis was that ongoing treatment over six months would maintain the lexical retrieval gains expected at the end of two or four weeks' treatment, and therefore lexical retrieval would be better at the end than the beginning of the study. We were not confident that we would see item generalisation because of the relative absence of this outcome in the PPA literature, but we designed our treatment to detect item generalisation if it should occur.