Verbal Short-Term Memory Disturbance in the Primary Progressive Aphasias: Challenges and Distinctions in a Clinical Setting

Impaired verbal ‘phonological’ short-term memory is considered a cardinal feature of the logopenic variant of primary progressive aphasia (lv-PPA) and is assumed to underpin most of the language deficits in this syndrome. Clinically, examination of verbal short-term memory in individuals presenting with PPA is common practice and serves two objectives: (i) to help understand the possible mechanisms underlying the patient’s language profile and (ii) to help differentiate lv-PPA from other PPA variants or from other dementia syndromes. Distinction between lv-PPA and the non-fluent variant of PPA (nfv-PPA), however, can be especially challenging due to overlapping language profiles and comparable psychometric performances on verbal short-term memory tests. Here, we present case vignettes of the three PPA variants (lv-PPA, nfv-PPA, and the semantic variant (sv-PPA)) and typical Alzheimer’s disease (AD). These vignettes provide a detailed description of the short-term and working memory profiles typically found in these patients and highlight how speech output and language comprehension deficits across the PPA variants differentially interfere with verbal memory performance. We demonstrate that a combination of verbal short-term and working memory measures provides crucial information regarding the cognitive mechanisms underlying language disturbances in PPA. In addition, we propose that analogous visuospatial span tasks are essential for the assessment of PPA as they measure memory capacity without language contamination.

NS was assessed following a 4-year history of speech and language difficulties. Initial symptoms included mispronouncing some words ("stunt" for 'stent'; "wiltered" for 'withered'), word substitutions, slowed reading rate, surface dyslexia, and spelling errors. Word-finding difficulties had reportedly become more apparent in the 2 years prior to Brain Sci. 2021, 11, 1060 4 of 21 the visit, particularly in instances that required rapid spontaneous speech. Cognitively, NS felt less confident about his memory and concentration. He also experienced some topographical disorientation in unfamiliar locations. No other cognitive or motor changes were reported, and he remained independent in all activities of daily living (ADLs). His wife had not noticed any behavioural or personality changes and there was no history of psychiatric features. NS did not report any symptoms of depression, anxiety, or stress, and he demonstrated appropriate emotional reactivity during the assessment.

Neuropsychological Assessment
Based on his educational and vocational history, NS's estimated premorbid level of functioning was average. On a measure of general cognitive ability, the Addenbrooke's Cognitive Examination-III (ACE-III), he scored 66/100 which was well-below normal limits (normal > 88), [13,25] (Table 1). His conversational speech was dysfluent with frequent word-finding pauses and phonological errors, though prosody was intact. Formal neuropsychological assessment revealed moderate to severe expressive language difficulties. Verbal fluency and confrontation naming were very impaired, and repetition of multisyllabic words was also reduced somewhat (Table 1, Figure 1). In contrast, comprehension (i.e., word-picture matching) and conceptual semantic knowledge were relatively preserved ( Figure 1). Very Mild; Word Span: word span test from Leyton, Savage [3]. * indicates borderline performance: 1.3 < z-score < 2.0; 3 < percentile < 9; ** indicates extremely low performance: z-score < −2.0; percentile < 2.
NS was assessed following a 4-year history of speech and language difficulties. Initial symptoms included mispronouncing some words ("stunt" for 'stent'; "wiltered" for 'withered'), word substitutions, slowed reading rate, surface dyslexia, and spelling errors. Word-finding difficulties had reportedly become more apparent in the 2 years prior to the visit, particularly in instances that required rapid spontaneous speech. Cognitively, NS felt less confident about his memory and concentration. He also experienced some topographical disorientation in unfamiliar locations. No other cognitive or motor changes were reported, and he remained independent in all activities of daily living (ADLs). His wife had not noticed any behavioural or personality changes and there was no history of psychiatric features. NS did not report any symptoms of depression, anxiety, or stress, and he demonstrated appropriate emotional reactivity during the assessment.

Neuropsychological Assessment
Based on his educational and vocational history, NS's estimated premorbid level of functioning was average. On a measure of general cognitive ability, the Addenbrooke's Cognitive Examination-III (ACE-III), he scored 66/100 which was well-below normal limits (normal > 88), [13,25] (Table 1). His conversational speech was dysfluent with frequent word-finding pauses and phonological errors, though prosody was intact. Formal neuropsychological assessment revealed moderate to severe expressive language difficulties. Verbal fluency and confrontation naming were very impaired, and repetition of multisyllabic words was also reduced somewhat (Table 1, Figure 1). In contrast, comprehension (i.e., word-picture matching) and conceptual semantic knowledge were relatively preserved ( Figure 1).

Figure 1.
Raw scores on the Naming, Repetition, and Comprehension subtests of the Sydney Language Battery. Note, due to speech output difficulties, the nfv-PPA patient wrote their responses for the Naming subtest (Appendix A). Verbal short-term memory (i.e., Digit Span Forward, Sentence Repetition, Word Span) and verbal working memory (i.e., Digit Span Backward) were extremely impaired (Tables 1 and 2, Figure 2). His performance on the visuospatial counterpart tasks was comparatively better although still fell in the borderline-impaired range (Table 1, Figure 3).
Executive functioning difficulties were also evident. Specifically, NS demonstrated impairments in proverb interpretation, and rapid set-shifting. Complex visuo-constructional planning was disorganised and extremely fragmented (Figure 4) in the context of intact basic visuo-perceptual skills and psychomotor speed. This was likely to have impacted on his visual memory performance which was borderline-impaired.  Take this home  1  0  0  0  0  0  2  Where is the child?  1  0  0  0  0  0  3  The car will not run  1  0  0  0  0  0  4  Why are they not living here?  1  0  0  0  0  0  5 The band (.) played and the/kraUnd/ (5) cheered  0  1  0  0  0  0  1  6 Where are you going to work next summer?
He sold his house/@n/moved to the farm 0 1 0 0 0 0 1 8 Work in the garden until you've picked all the beans 1 0 0 0 0 0 9 The artist/peIt@d/painted (3) many pictures of the/fA:/no sorry This doctor doesn't go to all of the towns She should be able to tell us when/S3:/(2) when she (.) is (.) performing 0 1 0 0 0 0 4 1 12 Why/d6n/(.) that group (1) why doesn't that group apply (.) for (.) money 0 1 0 0 0 0 8 1 * 1 13 Many/pi:p@l/(5) they were not able to get work because of the (.) weather 0 1 0 0 0 0 6 2 * 1 14 Did not attempt     He should actually be able to tell us when she will actually be performing here Why do members of that group never write to their representatives of their group for aid?
Many men and women were not able to get to their work because of the severe snowstorm Notes, Sentences were transcribed using the international phonetic transcription (IPA). Sentences were scored using the Hohlbaum, Dressel [12] scoring criteria. Notes, P: glottal stop; (2) represents pause in seconds; (.) represents pause of <1 s; *: These errors are coded as semantic substitutions but in fact they do not appear to arise as a result of impaired lexical retrieval. Rather the lv-PPA patient decodes the meaning but cannot repeat the content word by word. So, he paraphrases, e.g., people for men and women; **: These errors are additions. The Hohlbaum system codes them as semantic substitutions, although there is no specific code for such errors. Verbal short-term memory (i.e., Digit Span Forward, Sentence Repetition, Word Span) and verbal working memory (i.e., Digit Span Backward) were extremely impaired (Tables 1 and 2, Figure 2). His performance on the visuospatial counterpart tasks was comparatively better although still fell in the borderline-impaired range (Table 1, Figure  3).  Overall performance on Digit Span was very impaired for the lv-PPA patient, borderline-impaired for the nfv-PPA patient, and average for the sv-PPA and AD patients. Overall performance on Spatial Span was borderline-impaired for the lv-PPA and AD patients, and average for the nfv-PPA and sv-PPA patients. Overall Digit Span was significantly (0.05) worse than Spatial Span for the nfv-PPA patient; with the reverse pattern (i.e., Spatial < Digits) found for the AD patient. There was no statistical difference between test modality for lv-PPA and sv-PPA patients (statistical thresholds were taken from Table F4 [Appendix F] of the WMS III Scoring Manual).
Executive functioning difficulties were also evident. Specifically, NS demonstrated impairments in proverb interpretation, and rapid set-shifting. Complex visuo-constructional planning was disorganised and extremely fragmented ( Figure 4) in the context of intact basic visuo-perceptual skills and psychomotor speed. This was likely to have impacted on his visual memory performance which was borderline-impaired. Verbal short-term memory (i.e., Digit Span Forward, Sentence Repetition, Word Span) and verbal working memory (i.e., Digit Span Backward) were extremely impaired (Tables 1 and 2, Figure 2). His performance on the visuospatial counterpart tasks was comparatively better although still fell in the borderline-impaired range (Table 1, Figure  3).  Overall performance on Digit Span was very impaired for the lv-PPA patient, borderline-impaired for the nfv-PPA patient, and average for the sv-PPA and AD patients. Overall performance on Spatial Span was borderline-impaired for the lv-PPA and AD patients, and average for the nfv-PPA and sv-PPA patients. Overall Digit Span was significantly (0.05) worse than Spatial Span for the nfv-PPA patient; with the reverse pattern (i.e., Spatial < Digits) found for the AD patient. There was no statistical difference between test modality for lv-PPA and sv-PPA patients (statistical thresholds were taken from Table F4 [Appendix F] of the WMS III Scoring Manual).
Executive functioning difficulties were also evident. Specifically, NS demonstrated impairments in proverb interpretation, and rapid set-shifting. Complex visuo-constructional planning was disorganised and extremely fragmented ( Figure 4) in the context of intact basic visuo-perceptual skills and psychomotor speed. This was likely to have impacted on his visual memory performance which was borderline-impaired. . Digit and Spatial Span total raw scores graphically represented as age-adjusted z-values. Overall performance on Digit Span was very impaired for the lv-PPA patient, borderline-impaired for the nfv-PPA patient, and average for the sv-PPA and AD patients. Overall performance on Spatial Span was borderline-impaired for the lv-PPA and AD patients, and average for the nfv-PPA and sv-PPA patients. Overall Digit Span was significantly (0.05) worse than Spatial Span for the nfv-PPA patient; with the reverse pattern (i.e., Spatial < Digits) found for the AD patient. There was no statistical difference between test modality for lv-PPA and sv-PPA patients (statistical thresholds were taken from

Clinical Opinion about the Patient's Verbal and Visuospatial Short-Term and Working Memory Profile
NS demonstrates the hallmark verbal short-term memory disorder characteristic of lv-PPA, evidenced by impaired digit span, word span, and sentence repetition, in the context of relatively spared repetition of high frequency (i.e., lower cognitive load) multisyllabic words. Qualitatively, NS's intact repetition of multisyllabic words but impaired sentence repetition suggests that the latter arises from difficulties accessing and rehearsing verbal information in mind-that is, a dissolution of the verbal short-term memory system, rather than from deficits in motor speech production (Table 2) [11,12]. His mild articulatory errors and adequate prosody during speech, with well-articulated sound substitutions and lack of distortions further supports this position. Notably, NS's short-term and working memory impairments appear to extend beyond the verbal domain, evidenced by his impaired visuospatial span. Brain Sci. 2021, 11, x FOR PEER REVIEW 6 of 23 Copy performance was very impaired for the AD patient, borderline-impaired for the lv-PPA and nfv-PPA patients, and within normal limits for the sv-PPA patient. Three-minute recall performance was very impaired for the AD patient, borderline-impaired for the lv-PPA patient, and within normal limits for the nfv-PPA and sv-PPA patients.

Clinical Opinion about the Patient's Verbal and Visuospatial Short-Term and Working Memory Profile
NS demonstrates the hallmark verbal short-term memory disorder characteristic of lv-PPA, evidenced by impaired digit span, word span, and sentence repetition, in the context of relatively spared repetition of high frequency (i.e., lower cognitive load) multisyllabic words. Qualitatively, NS's intact repetition of multisyllabic words but impaired sentence repetition suggests that the latter arises from difficulties accessing and rehearsing verbal information in mind-that is, a dissolution of the verbal short-term memory system, rather than from deficits in motor speech production (Table 2) [11,12]. His mild articulatory errors and adequate prosody during speech, with well-articulated sound substitutions and lack of distortions further supports this position. Notably, NS's short-term and working memory impairments appear to extend beyond the verbal domain, evidenced by his impaired visuospatial span. Copy performance was very impaired for the AD patient, borderline-impaired for the lv-PPA and nfv-PPA patients, and within normal limits for the sv-PPA patient. Three-minute recall performance was very impaired for the AD patient, borderline-impaired for the lv-PPA patient, and within normal limits for the nfv-PPA and sv-PPA patients.

Brain MRI and Clinical Diagnosis
T1 coronal brain MR images revealed mild generalised cortical atrophy, slightly more prominent on the left than the right, extending posteriorly to involve the parietal lobes ( Figure 5). There was marginally greater atrophy in the peri-insular region on the left than the right. T2 weighted MRI images showed occasional hyperintensities in the cerebral hemispheres which were within normal limits for NS's age. The pattern of brain atrophy, clinical history, language, and neuropsychological profile were consistent with a diagnosis of lv-PPA (Table 3a). Pathological confirmation was unavailable as NS is still alive.

nfv-PPA Patient: ML
At presentation, ML was a 64-year-old, right-handed woman with 12 years of education (Table 1). She had been retired for five years, having previously worked as a shop owner and a public servant. She had a past history of liver disease due to hepatitis C, hepatic cirrhosis, and long-term alcohol consumption. At the time of the assessment, ML had been abstinent from alcohol for 7 years. Her father had been diagnosed with dementia (type unknown) in his 80s and died at the age of 87.
ML was seen following a 5-year history of progressively deteriorating speech which had worsened noticeably over the last 12 months. Initial symptoms also included frequent spelling errors and incorrect sentence construction. Her husband reported that her text messages often had errors but remained largely understandable. She reported occasional Yes/No and Hi/Bye confusion but had no trouble using corresponding non-verbal gestures. No other cognitive changes were reported. She described no swallowing difficulties and no Parkinsonian symptoms. She did not report any symptoms of depression, anxiety, or stress on a self-report measure of recent mood, and she demonstrated appropriate emotional reactivity during the assessment. According to her husband, her ADLs were mildly impaired. T1 coronal brain MR images revealed mild generalised cortical atrophy, slightly more prominent on the left than the right, extending posteriorly to involve the parietal lobes ( Figure 5). There was marginally greater atrophy in the peri-insular region on the left than the right. T2 weighted MRI images showed occasional hyperintensities in the cerebral hemispheres which were within normal limits for NS's age. The pattern of brain atrophy, clinical history, language, and neuropsychological profile were consistent with a diagnosis of lv-PPA (Table 3a). Pathological confirmation was unavailable as NS is still alive.

Figure 5.
Brain T1 magnetic resonance images (MRI) of patients at the time of neuropsychological testing. Note: L = left; R = right. Brain images are presented in the coronal plane. lv-PPA patient: Mild generalised cortical atrophy was evident, slightly more prominent on the left than the right, extending posteriorly to involve the parietal lobes. There was marginally greater atrophy in the peri-insular region on the left than the right. nfv-PPA patient: Mild generalised cortical atrophy with particular involvement of the left peri-insular region anteriorly. sv-PPA patient: Severe atrophy of the anterior temporal pole bilaterally, but much worse on the left than the right. AD patient: Mild-moderate generalised cortical atrophy with involvement of the mesial temporal lobes. Brain images are presented in the coronal plane. lv-PPA patient: Mild generalised cortical atrophy was evident, slightly more prominent on the left than the right, extending posteriorly to involve the parietal lobes. There was marginally greater atrophy in the peri-insular region on the left than the right. nfv-PPA patient: Mild generalised cortical atrophy with particular involvement of the left peri-insular region anteriorly. sv-PPA patient: Severe atrophy of the anterior temporal pole bilaterally, but much worse on the left than the right. AD patient: Mild-moderate generalised cortical atrophy with involvement of the mesial temporal lobes.

Neuropsychological Assessment
Based on her educational and vocational history, ML's premorbid intellectual ability was estimated to be average. She scored 81/100 on the ACE-III, which was below established cut-off scores for normal performance (normal > 88; Table 1). Formal neuropsychological testing revealed a primary impairment in expressive language. Qualitatively, her speech was markedly dysfluent, and contained articulatory and occasional grammatical errors. Single-word repetition and verbal fluency were extremely impaired on testing. Her other language abilities (comprehension, semantic knowledge) and writing, however, remained preserved (Table 1, Figure 1). Indeed, she could provide a reliable history of her difficulties by writing her responses. Performance on verbal short-term and working memory measures were also reduced, but likely due to her dysfluent speech (Table 2, Figure 2). Visuospatial short-term and working memory, on the other hand, was sound ( Figure 3). Encoding and retention of verbal and visual information was preserved. While basic visuo-perceptual abilities were intact, ML had subtle visuo-constructional difficulties evidenced by an imprecise and slightly disorganised copy of the Rey Complex Figure  (Figure 4). Other aspects of executive functioning (rapid set-shifting, inhibitory control) as well as psychomotor speed were impaired.

Clinical Opinion about the Patient's Verbal and Visuospatial Short-Term and Working Memory Profile
ML's profile is characteristic of nfv-PPA. Her verbal short-term and working memory span and sentence repetition were markedly impaired; however, frank motor speech deficits largely contributed to her impaired performance (Tables 1 and 2). Notably, ML had greater difficulty with repeating sentences and multisyllabic words or phrases than (predominantly monosyllabic) digits. In contrast, her visuospatial short-term and working memory performance appeared relatively intact (Figure 3).

Brain MRI and Clinical Diagnosis
The T1 coronal brain MR images revealed mild generalised cortical atrophy with particular involvement of the left peri-insular region anteriorly ( Figure 5). Cerebral atrophy over the convexity was also present with widening of the interhemispheric fissure. The pattern of brain atrophy, clinical history, language, and neuropsychological profile were consistent with a diagnosis of nfv-PPA (Table 3b). Pathological confirmation was unavailable as ML remains alive.

sv-PPA Patient: JC
At presentation, JC was a 62-year-old, right-handed man (Table 1). He completed 16 years of education and worked as a principal of a primary school before retiring 2 years prior to the visit. His medical history was significant for a parathyroid cancer 13 years prior which was treated with a thyroidectomy and subsequently managed with levothyroxine, and ischaemic heart disease, with a myocardial infarction and 4-vessel coronary artery bypass graft surgery 4 years prior to the assessment. At the time of the assessment, JC was on antiplatelet and cholesterol-lowering medications. There was no report of symptoms of depression, anxiety, or stress on a self-report measure of recent mood and there were no significant periods of mood disorder noted. He demonstrated appropriate emotional reactivity during the assessment. There was no known family history of dementia.
JC was assessed following a 6-year history of speech and language difficulties with initial symptoms including forgetfulness and difficulties learning students' names at work. He reported progressive difficulties recalling names of people and objects (i.e., plants, animals) in the past 3 years, as well as a decline in his language comprehension and semantic knowledge. He was reportedly an avid reader previously though this had declined due to difficulties understanding the meaning of words. No issues with reading the actual words or recognising letters were reported. According to JC's wife, there was no change in behaviour or personality, eating habits, appetite, or weight. JC remained physically well and had no weakness or motor dysfunction. According to his wife, his ADLs were mildly impaired.

Neuropsychological Assessment
Based on his education and vocational history, JC's premorbid intellectual ability was estimated to lie within the average to high average range. He scored 67/100 on the ACE-III which was well below normal limits (normal > 88, Table 1). Qualitatively, his speech output was fluent with no phonological errors or substitutions, though occasional word-finding problems were noted, and he was slightly circumlocutory. Formal neuropsychological testing revealed intact visuospatial short-term and working memory (Table 1, Figure 3). Whilst his verbal working memory for numerical information and single words was intact, sentence repetition was compromised ( Figure 2, Table 2). In terms of his language, JC demonstrated impaired confrontation naming, single-word comprehension and semantic knowledge; single-word repetition and verbal fluency, however, remained intact ( Figure 1). Aside from suboptimal set-shifting, no significant executive functioning impairments were evident. JC's visuo-constructional planning and organisation remained intact. No visual memory deficits were apparent ( Figure 4); detailed assessment of JC's verbal learning and memory, however, was not conducted on this occasion.

Clinical Opinion about the Patient's Verbal and Visuospatial Short-Term and Working Memory Profile
Overall, JC's cognitive profile is consistent with the characteristic sv-PPA profile. Whilst his short-term and working memory for digits, words and visuospatial information remained relatively spared, his repetition of complex sentences was more problematic. Notably and consistent with his intact basic verbal short-term memory span, qualitative appraisal of his sentence repetition (Table 2) suggests that his poor performance results from inclusions of grammatically correct but superfluous words (see Discussion).

Brain MRI and Clinical Diagnosis
T1 coronal brain MR images showed bilateral atrophy in the temporal lobes (left more markedly than right) particularly affecting the left hippocampal region and periinsular region. Mild atrophy of the frontal lobes bilaterally was also evident ( Figure 5). T2 weighted MRI images revealed evidence of scattered white matter hyperintensities in both hemispheres in keeping with small vessel ischaemic change. No established territorial infarcts were evident. The pattern of brain atrophy, clinical history, language, and neuropsychological profile were in keeping with a diagnosis of sv-PPA (Table 3c). JC underwent a Pittsburgh compound B (PiB) positron emission tomography scan (PiB-PET), which uses a radio-ligand of amyloid protein as a biomarker for AD [26]. The patient showed a low uptake of the PiB tracer, suggesting the absence of underlying Alzheimer pathology [26].

Typical AD Patient: TN
TN presented as a right-handed 67-year-old male. He completed nine years of formal education and worked in the government services for three years before working in hospitality and owning a business. At the time of assessment, he had been retired for 4 years. His medical history included ischaemic heart disease, with coronary artery bypass graft surgery 19 years prior to the assessment. TN also had a history of post-traumatic stress disorder arising from his previous employment, though this was well-managed at the time of assessment. He did not report significant symptoms of depression, anxiety, or stress on a self-report measure of recent mood. Regular medications included telmisartan for hypertension, antiplatelet medication, and selective serotonin reuptake inhibitors (SSRIs) for depressive symptoms. There was no known family history of neurodegenerative disease.
TN presented for assessment following a 7-year history of insidious memory decline, which had worsened considerably in the last 3 years. Both TN and his wife reported difficulties with his memory, specifically with names and topographical memory. Some organisational and planning difficulties were also noted which impacted his instrumental ADLs. TN's wife had noticed mild apathy but there were otherwise no other personality or behavioural changes.

Neuropsychological Assessment
Based on his education and vocational history, TN's estimated premorbid intellectual ability fell in the average range. He scored 68/100 on the ACE-III, which was well below normal limits (normal > 88; Table 1). Consistent with his diagnosis, TN demonstrated prominent verbal and visual memory impairment on testing. He was unable to learn a word list over repeated trials, and his recall of a previously copied two-dimensional complex geometric figure after a 3-minute delay was extremely poor (Figure 4). Whilst verbal short-term and working memory was intact, visuospatial short-term memory was reduced, and his visuospatial working memory (i.e., Spatial Span Backward) was borderlineimpaired (Table 1). Other executive functioning abilities were variable-he was extremely slow and made several errors on a set-shifting task. Basic visuo-perceptual skills and visuo-constructional abilities (e.g., drawing simple objects) were preserved, although some higher-level visuo-constructional difficulties were present (Figure 4). Psychomotor speed was intact. Finally, no overt expressive language issues were noted in conversation although confrontation naming and comprehension (i.e., word-picture matching) were below expectations on testing (Figure 1). Other aspects of language (repetition, verbal fluency, and higher-level semantic knowledge) were relatively intact.

Clinical Opinion about the Patient's Verbal and Visuospatial Short-Term and Working Memory Profile
TN's verbal and visuospatial span profile was consistent with typical AD (Figure 3). As expected from a typical AD diagnosis, TN's visuospatial working memory ability (as demonstrated on Spatial Span Backward) was poor, whilst basic verbal and visuospatial short-term memory span remained relatively preserved.

Brain MRI and Clinical Diagnosis
T1 coronal brain MR images revealed atrophy of the medial temporal region bilaterally including the hippocampus, as well as diffuse frontal and parietal atrophy ( Figure 5). Ventricular enlargement was also evident. T2 weighted MRI images revealed scattered white matter hyperintensities in both hemispheres in keeping with small vessel ischaemic change. No established territorial infarcts were evident.
Two years after his assessment, TN underwent in vivo amyloid-PET imaging, which showed uptake of the amyloid ligand above the cut-off for an amyloid based pathology, indicating the presence of underlying Alzheimer disease [26]. The clinical history, language, neuropsychological profile, and confirmation of underlying Alzheimer pathology were consistent with a clinical diagnosis of typical AD [27].

Summary of the Short-Term and Working Memory Profiles across Patients and Relative to a Matched Control Group
These short-term and working memory profiles were established based on standardised norms from various population groups and sizes: WAIS-III and WMS-III Digit and Spatial Span [17,22], and Word Span and Sentence Repetition norms [4]. To ensure that our findings were not due to differences across normative populations, performance profiles of the case vignettes were compared to one sample of Australian matched controls [4] and are displayed as percentage scores ( Figure 6).
Inspection of these scores confirm that the lv-PPA and nfv-PPA patients were disproportionately impaired on Digit and Word Span relative to Spatial Span and overall cognitive ability (i.e., ACE total). By contrast, performance differences across verbal and visuospatial modalities were less evident for the sv-PPA and AD patients. Sentence Repetition performance across patients and within individual performance profiles was variable and uninterpretable across patients based on raw scores alone.
Regarding the specific short-term and working memory profiles, the lv-PPA, sv-PPA and AD patients demonstrated greater impairment on the Digit Span Backward than Forward tasks compared to controls. The reverse pattern was found for the nfv-PPA patient. These findings suggest that, over and above the inherent general difficulty associated with verbal working memory, the lv-PPA, sv-PPA and AD patients displayed more difficulty on this task relative to their verbal short-term memory capacity. As previously discussed, the nfv-PPA patient's impoverished speech is likely to have contributed to their performance on this task. Inspection of these scores confirm that the lv-PPA and nfv-PPA patients were disproportionately impaired on Digit and Word Span relative to Spatial Span and overall cognitive ability (i.e., ACE total). By contrast, performance differences across verbal and visuospatial modalities were less evident for the sv-PPA and AD patients. Sentence Repetition performance across patients and within individual performance profiles was variable and uninterpretable across patients based on raw scores alone. Overall Spatial Span profiles were distinct across patients. Relative to overall cognitive ability (i.e., ACE Total), the sv-PPA and nfv-PPA patients' Spatial Span performance was relatively spared. By contrast, the lv-PPA and AD patient's Spatial Span performance was reduced somewhat. These findings were in keeping with other measures of visuospatial episodic memory (i.e., RCFT 3-minute recall) and visuo-construction (RCFT Copy).

Discussion
While significant advancements have been made in the classification of PPA and its variants, challenges remain in clinical practice in differentiating PPA profiles in individual patients. Using case vignettes, we demonstrate that the canonical phonological disturbance displayed in lv-PPA is distinguishable from other PPA language profiles when using a selection of basic language and short-term memory measures. Importantly, we highlight how speech output and language deficits differentially interfere with verbal memory performance across the PPA variants, and how these differences provide insights into the underlying cognitive processes affected in these syndromes. Further, we demonstrate that visuospatial span tasks are essential for the assessment of PPA as they measure memory capacity without language contamination.
The lv-PPA patient demonstrates the typical verbal short-term memory deficit observed in this syndrome: very impaired digit and word span, and markedly reduced sentence repetition, in the context of relatively spared repetition of single two-and threesyllabic words [1,2]. Importantly, this occurred in the absence of frank motor speech deficits. Phonological paraphasias were, however, present. In contrast the nfv-PPA patient also performed poorly on verbal repetition tasks, but this performance was contaminated by frank motor speech deficits. Notably, and consistent with previous studies, we found that the nfv-PPA patient's verbal short-term memory performance declined as the motor speech sequencing requirements increased (i.e., repeating monosyllabic digits compared to a span of multisyllabic words and/or phrases [3]). Considering the sv-PPA patient's profile, we found digit and word span were spared whereas sentence repetition was compromised, likely to be due to his degraded semantic store [3]. Specifically, it is thought that the dissolving semantic knowledge and ability to form conceptual representations in sv-PPA may impact on the capacity to 'chunk' verbal material into meaningful components-a skill necessary for holding larger quantities of verbal information [3,11,28]. While not systematically verified in the current study, the sv-PPA patient's occasional circumlocutory responses and/or word inclusions on the Sentence Repetition task would support this interpretation.
Taken together, we propose that, when assessing for lv-PPA, Digit Span Forward and Word Span tasks are more robust measures of verbal short-term memory than Sentence Repetition or Digit Span Backward tasks, as the former tasks are less susceptible to language and/or dysexecutive contamination [4,11,12]. This observation notwithstanding, important qualitative information can be gained from Sentence Repetition that will help with the distinction between the two non-fluent PPA syndromes; specifically, its ability to elicit phonological paraphasias (in lv-PPA) or motor speech deficits (in nfv-PPA) [11,12]. For the assessment of sv-PPA, we caution that some patients may perform poorer on Sentence Repetition than other span tests as their degraded semantic knowledge may preclude their ability to 'chunk' and/or form meaningful representations in mind [3,11,28].
In contrast to their similar verbal span and sentence repetition performance, the lv-PPA and nfv-PPA patients showed distinct visuospatial span profiles. The disproportionately compromised visuospatial short-term and working memory in lv-PPA relative to the other PPA variants is consistent with a growing body of research and alludes to their distinct neuroanatomical profile [4,[29][30][31][32]. Briefly, brain regions involved in visuospatial short-term and working memory, including bilateral posterior temporal and parietal brain structures, are more compromised in lv-PPA and AD than in nfv-PPA and sv-PPA [4,[33][34][35]. Undoubtedly, and unlike in nfv-PPA where the deficit remains primarily verbal, the impaired short-term memory of lv-PPA extends to the non-verbal domain, even at low levels of difficulty [30,31,36,37]. Based on these findings, we propose that when assessing PPA patients without motor features, difficulty on basic/lower-load visuospatial short-term memory tests strongly indicates lv-PPA [31,38,39]. This distinction will assist the diagnostic process, particularly in the presence of a mixed language profile.
It is notable that short-term and working memory deficits are not limited to PPA. Indeed, we found that the AD patient showed impairments across these domains. Unlike the PPA patients, however, the AD patient's overall visuospatial span was significantly worse than their overall verbal span profile. These findings are consistent with the commonly held view that multiple components of visuospatial memory, including processing, integration, storing, and retaining visual material, break down in the earlier stages of typical AD [4,31,33,40]. Consistent with this view, the visuospatial difficulties of the AD patient extended beyond Spatial Span-with deficits also noted on complex visuo-constructional, visuospatial episodic memory, and attentional tasks (i.e., Rey Complex Figure, Trails A  and B). Importantly, the widespread memory and visuospatial deficits of the AD patient supports the opinion that AD is distinct from lv-PPA [31,41]. That is, lv-PPA is not simply typical AD with additional language deficits [42]. While this nuanced distinction may seem unnecessary, it has clinical implications when addressing the care needs, treatment options, and estimated survival of either AD or lv-PPA [43]. To that end, we propose requisite assessment of both verbal and visuospatial cognitive domains for the differential diagnosis of AD and lv-PPA.
In clinical practice, awareness that the cognitive profiles of PPA and AD vary across individuals is vital. Most studies compare matched PPA subgroups at a single point in time (typically at the mild to moderate disease severity stage) and provide findings which typically overemphasise the differences across variants but underemphasise the differences within each syndrome. Studies that have investigated within syndromes, however, suggest that the language and cognitive profile of lv-PPA varies considerably [6][7][8][9][10]44]. For example, it is reported that while most lv-PPA patients present with multi-domain cognitive impairment at baseline assessment, a subset of lv-PPA patients present with relatively circumscribed language deficits with relatively mild cognitive deficits in other domains [6]. Important to this topic is the awareness that, with disease progression, language and cognitive abilities of PPA and AD inevitably decline, eventuating in a manifold dissolution of functional abilities [45][46][47]. As such, the distinct cognitive profiles observed in the earlier stages of these diseases may become less apparent in later disease stages. To illustrate this point, Table 4 demonstrates the decline of Spatial and Digit Span performance across PPA and AD patients, stratified by overall cognitive ability. Put together, clinicians should take a gestalt approach to assessing PPA and AD in clinical practice and consider the 'moving parts' of language and cognitive deficits, as well as overall cognitive ability, before forming a formal clinical diagnosis.
Several caveats warrant attention. Pathological confirmation was not available in any of the PPA cases. Nonetheless, the clinical, cognitive, and imaging information provided in the current study is typical of what is commonly available at baseline assessment in routine clinical practice (i.e., non-tertiary/specialised centres) and was sufficient to establish a clinical diagnosis of each PPA variant [1]. It is reported that~70% of lv-PPA patients have underlying Alzheimer pathology,~70% of nfv-PPA have tauopathy, and~85% of sv-PPA have TDP−43, and that detailed and careful clinical, cognitive, and imaging examination improves this pathological correspondence [48][49][50]. More research, however, is warranted to determine if specific cognitive profiles within PPA syndromes can further discern the pathological course. The nfv-PPA patient in this study demonstrated a severe motor speech disorder with severe articulatory and prosodic impairment. We are aware, however, that other nfv-PPA language profiles exist (e.g., agrammatism and articulatory impairment without motor speech problems). Of particular interest to our team is the extent to which nfv-PPA language profiles without motor speech problems interfere with verbal short-term memory measures. Future research is needed to determine the extent to which verbal short-term memory performance is compromised across the distinct nfv-PPA language profiles.
Lastly, a proportion of nfv-PPA patients will develop Parkinsonian features (i.e., limb apraxia, akinesia/bradykinesia, motor rigidity) as the disease progresses (typically in the moderate to severe disease stages) [51]. In the current study, we do not refer to these patients as we assumed that they are more easily distinguishable from lv-PPA based on their clinical profiles alone (as Parkinsonian features are not common in lv-PPA) [51]. We acknowledge, however, that nfv-PPA patients with Parkinsonian features are likely to perform poorly on visuospatial related tasks due to their inherent motor dysfunction [38,39]. We therefore advise that our findings are only applicable to nfv-PPA without Parkinsonian features. Future research is warranted to delineate the nfv-PPA visuospatial short-term and working memory profiles with or without concomitant Parkinsonian features.

Conclusions
In summary, using case vignettes, we demonstrate the canonical verbal short-term memory profile of lv-PPA and how it differs from the other PPA variants as well as typical AD. Importantly, we demonstrate that a combination of verbal short-term and working memory measures commonly used in clinical settings can provide crucial information regarding the cognitive mechanisms underlying language disturbances across PPA variants. Further, we demonstrate that visuospatial span tasks are essential for the assessment of PPA as they measure memory capacity without contamination of language ability.  Data Availability Statement: The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. No part of the study procedures or analyses were preregistered prior to the research being undertaken. The Addenbrooke's Cognitive Examination-Third edition (ACE-III) and Sydney Language Battery (SYDBAT) are freely available at https://frontierftd.org (accessed on 11 August 2021). Legal copyright restrictions prevent public archiving of the other neuropsychological tests used in this research. These materials can be obtained from the copyright holders in the cited references.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
Administration and scoring guidelines of the SYDBAT can be found at https:// frontierftd.org (accessed on 11 August 2021). For the administration of the Naming subtest, the examiner is required to display the item/picture and ask the participant "What is this called?". The participant is encouraged to give their response in a timely manner but is not penalised for a delayed response. In accordance with the SYDBAT administration guidelines, the examiner is required to prompt the patient in these circumstances: (i) if the participant provides an abbreviated version of the word (e.g., "bike", "PC"), the examiner should say "What is the full name?"; (ii) if the participant provides a response that is vague or describes the item (e.g., "oh that's a type of food"), the examiner should say "Can you give me the exact name?"; (iii) if the participant uses an alternative or colloquial version (e.g., "wireless" rather than "radio" for Item 8), the examiner should say "Do you know another word for that?" or "What is the formal name for that?"; (iv) if the participant grossly misperceives an item, the examiner should clarify it. For example, for Item 25 (Tiara), some participants do not recognise that it sits on top of someone's head. The examiner should clarify and say, "This is on someone's head". For another example, for item 28 (balaclava), if the participant responds with "robber", the examiner should say "But what is the name of the thing he is wearing?". No other prompts are permitted for the SYDBAT Naming subtest. In this case study, the sv-PPA patient received two (iv) prompts, and the lv-PPA and nfv-PPA patient received one (i) prompt. It is strongly encouraged that participants provide a spoken/verbal response for the Naming subtest. Given the extent of the nfv-PPA patient's speech output problems, however, it was decided that it was more clinically useful to test her ability to freely-recall the names of items without spoken language (i.e., in written form). To remain consistent with the scoring guidelines for spoken responses, the nfv-PPA patient was penalised for spelling errors. Written responses were also permitted for the ACE-III Orientation, Attention, Memory, and Naming items. For all other items requiring speech (i.e., ACE-III: Verbal Fluency, Repetition, Reading; Digit Span; Letter Fluency; Sentence Repetition; SYDBAT Repetition etc.), she was required to provide spoken/verbal responses.