Throughout the progression of primary progressive aphasia (PPA), research supports the use of behavioral interventions targeting two complementary goals: (1) reducing impairments through restitutive therapies¹ enhancing communication, and (2) increasing daily life participation² through more compensatory approaches (Robinaugh and Henry, 2022; Taylor-Rubin et al., 2021). Restitutive therapies aim to restore lost functions and are typically adapted from well-established therapies for stroke-induced aphasia that target language skills, such as word finding, comprehension, and sentence production. Compensatory approaches focus on supporting individuals in finding alternative ways to accomplish tasks and enhance their active participation despite communication impairments.

Although therapies for PPA resemble those for stroke-induced aphasia, outcomes must be interpreted in light of progressive decline. Unlike stroke-induced aphasia, which results from a sudden, delimited brain injury and exhibits spontaneous recovery, PPA presents distinct challenges with its progressive language decline due to more diffuse neurodegeneration, even when receiving therapy (Norise and Hamilton, 2016). Therefore, a necessary consideration when measuring therapy efficacy in individuals with PPA is the expected gradual decline in abilities throughout the intervention period. Still, it remains crucial to assess the retention of skills present at baseline and the learning of newly acquired skills after the end of therapy (maintenance) and their generalization to other target words and language abilities (Cotelli et al., 2020). These measures are essential for understanding the functional significance of a given therapy and its impact on participants' daily communication. See Cotelli et al. (2020), Beales et al. (2018), and Cadorio et al. (2017) for reviews on maintenance and generalization following naming treatments with individuals living with PPA.

A recent systematic review identified 103 studies that reported behavioral treatments for individuals living with PPA (Wauters et al., 2023). Many studies were focused on restitutive therapies targeting anomia and agraphia since these are among the most common deficits in all variants. Anomia, for example, is a symptom encountered in all PPA variants. However, it may be caused by a breakdown at different levels of the naming process (e.g., at the conceptual level in individuals with semantic variant PPA (svPPA) and the lexical access level in individuals with logopenic PPA (lvPPA). Few studies included in the review investigated restitutive therapy targeting other areas of language and communication, such as semantic knowledge, phonology, speech production and fluency, syntax and morphology, word comprehension, spoken discourse production, or general language and cognition, and compensatory therapies such as augmentative and alternative communication and more functional communication strategies (Wauters et al., 2023).

Restitutive therapies targeting anomia in individuals living with PPA have yielded encouraging results immediately following intervention for trained items and generalization to untrained items, as well as leading to maintenance of therapy gains several weeks after therapy. Indeed, reviews supporting the benefits of spoken naming therapies have been published in recent years (Robinaugh and Henry, 2022; Taylor-Rubin et al., 2021; Wauters et al., 2023; Volkmer et al., 2020). Various approaches have been documented with individuals living with PPA, including semantic therapies emphasizing the meaning of the target words and phonological/orthographic therapies involving sound or spelling cues for the target words, for both nouns and verbs (Robinaugh and Henry, 2022; Taylor-Rubin et al., 2021; Wauters et al., 2023; Volkmer et al., 2020). These therapies enhance semantic or phonological activation to support word retrieval, following psycholinguistic models like the restricted interaction account (Goldrick and Rapp, 2002). Although maintenance and generalization are not measured in all studies, most of the high-quality studies identified in the review by Wauters et al. (2023) documented at least one participant exhibiting generalization and maintenance of treatment gains. The magnitude of therapy effects in these studies was small, as could be expected given the context of a neurodegenerative disorder.

Furthermore, restitutive behavioral therapies targeting agraphia in individuals living with PPA have also shown positive results as reported by recent reviews (Taylor-Rubin et al., 2021; Wauters et al., 2023), and more specifically, work done by our team (Rapp and Glucroft, 2009; Tsapkini and Hillis, 2013; Tsapkini et al., 2014, 2018). In an initial study, Tsapkini et al. (2014) implemented an intervention targeting the relearning of phoneme-to-grapheme conversion (PGC, also known as sound-to-letter correspondence), a skill that is often impaired in individuals with PPA. This intervention yielded gains for the trained phoneme-grapheme pairs (Tsapkini et al., 2014). In a subsequent trial, Tsapkini et al. (2018) applied the “spell-study-spell” therapy, representing a “lexical” therapy focusing on training the spelling of entire words (Rapp and Glucroft, 2009). Again, therapy led to improvements in spelling for the trained words; however, there was no generalization to untrained words when therapy was given alone (Tsapkini et al., 2018).

Neuromodulation methods, like tDCS, have attracted significant interest over the past decade as promising adjunctive interventions for individuals living with PPA. As recent studies and meta-analyses of combined behavior and neuromodulation interventions in individuals with PPA have shown, a great strength of these approaches (and tDCS in particular) is the increased maintenance and generalization effects compared to behavioral therapy alone, at least in spoken naming and spelling (Cotelli et al., 2020; Coemans et al., 2021; Wang et al., 2023; Nissim et al., 2020). tDCS involves safe, non-invasive application of electrical currents to modulate neuronal activity in targeted brain regions, enhancing neural plasticity (Norise and Hamilton, 2016; Rahman et al., 2013; Sebastian et al., 2016; Stagg and Nitsche, 2011) and potentially aiding language recovery in individuals with PPA by modulating cortical excitability (Ficek et al., 2018; Meinzer et al., 2015; Tao et al., 2021).

tDCS is thought to modulate neural activity by altering cortical excitability and neurotransmitter concentrations, particularly by subthreshold modulation of neuronal membranes (Stagg and Nitsche, 2011). In PPA, tDCS has been shown to modulate functional connectivity (FC), which can serve as an indirect correlate of cortical excitability. In a similar crossover, double-blinded, sham-controlled trial preceding the present one, members of our team found that a-tDCS reduced hyperconnectivity between the stimulated left IFG and functionally connected areas more than sham-tDCS, both paired with the same oral and written naming therapy. Furthermore, these reductions correlated with improved language outcomes up to 2 months post-treatment (Ficek et al., 2018; Tao et al., 2021). These decreases usually reflect increased task efficiency as behavior becomes more automated and have been linked to better treatment outcomes (Meinzer et al., 2015). Notably, both reductions and increases in FC have been observed depending on baseline connectivity patterns, suggesting that tDCS may normalize by downregulating or upregulating either hyper- or hypoconnectivity, respectively, depending on the baseline, to support language recovery (Tao et al., 2023). Additionally, reductions in the inhibitory neurotransmitter GABA following tDCS have also been linked to behavioral gains in individuals with PPA (Harris A. D. et al., 2019). Emerging evidence points to variant-specific changes in FC and metabolic alterations induced by tDCS, as measured by FDG PET, extending mechanistic understanding of how tDCS supports recovery in PPA (Tao et al., 2023; Roncero et al., 2023).

Compared to other neuromodulation methods, such as transcranial magnetic stimulation (TMS), tDCS is significantly low-cost and easily portable, making it attractive to implement in home-based therapies (Tippett et al., 2024; Im et al., 2019; Cappon et al., 2024), a feasible and efficacious approach in individuals with PPA, in particular (Neophytou et al., 2024). The simultaneous use of tDCS with behavioral language therapies aligns with the current understanding of neuromodulation mechanisms: the effects of stimulation on cortical reorganization appear to be influenced by the individual's state or current effort, with behavioral activity during or near the time of stimulation directly impacting outcomes (Silvanto et al., 2008; Gill et al., 2015). Research suggests that maximum benefits from stimulation are obtained when the brain area is actively engaged in a relevant behavioral task during or near the time of stimulation (Bikson and Rahman, 2013). The specificity of tDCS has also been demonstrated in PPA, as generalization of tDCS effects occurred only for functions related to the stimulated area (Wang et al., 2023).

Different PPA variants are associated with distinct patterns of neurodegeneration that affect language processing. For example, individuals living with lvPPA typically show predominant left posterior perisylvian or parietal atrophy, while the neurodegeneration in individuals living with non-fluent variant PPA (nfvPPA) is marked by predominant left posterior fronto-insular atrophy (Gorno-Tempini et al., 2011). These regions are critically involved in phonological working memory, syntactic processing, and motor speech planning. Individuals living with svPPA display distinct predominant anterior temporal lobe atrophy and semantic impairments (Gorno-Tempini et al., 2011). This study focuses on lvPPA and nfvPPA, where stimulation of the left inferior frontal gyrus (IFG) is theoretically and anatomically relevant. Given the importance of the left IFG across various language functions, this area has become a key target for neuromodulation in tDCS studies aimed at enhancing language outcomes in PPA (Cotelli et al., 2020; Nissim et al., 2020).

For over 10 years, the research teams involved in this multisite study have investigated the potential of tDCS as a catalyst for therapeutic interventions. In the same studies discussed previously, Tsapkini et al. (2014, 2018) investigated the effectiveness of tDCS paired with speech-language treatment in individuals with PPA through randomized, double-blind, sham-controlled, within-subject crossover designs. In these studies, anodal-tDCS (a-tDCS) was applied over the left IFG simultaneously with the written naming/spelling therapies (Tsapkini et al., 2014; 2018). These studies produced positive outcomes with a-tDCS combined with behavioral therapy, yielding over 15% more improvement than behavioral therapy paired with sham-tDCS (s-tDCS) for individuals with PPA, albeit at different degrees for each variant. Significantly, a-tDCS promoted the generalization of treatment effects to untrained items and beyond the targeted tasks to functions subserved by the stimulated area (Tsapkini et al., 2018; Wang et al., 2023). Studies by Roncero et al. (2017, 2019) have found similar results following left inferior parieto-temporal stimulation combined with spoken naming therapy: a-tDCS, when combined with behavioral therapy, led to over 20% more improvement than s-tDCS both paired with the same behavioral therapy, and to the generalization of treatment effects to untrained items. Other research groups have also reported that a-tDCS interventions lead to greater maintenance of gains over time compared to behavioral therapies alone, highlighting the potential of this non-invasive neuromodulation technique as an adjunct to behavioral therapies in the management of PPA. As research in this area continues to evolve, tDCS offers new possibilities for improving language outcomes and quality of life in individuals living with PPA (Cotelli et al., 2020; Coemans et al., 2021; Nissim et al., 2020; Sebastian et al., 2016; Byeon, 2020).

Predicting therapy outcomes in individuals living with PPA remains a complex endeavor, with numerous factors influencing treatment response. Factors such as the PPA variant have been shown to influence therapy response. For example, it was found that those with non-fluent variant PPA (nfvPPA) tend to benefit more from tDCS (i.e., improve in both trained and untrained material) (Tsapkini et al., 2018). Moreover, individuals living with either nfvPPA or lvPPA have shown good therapy response and generalization (Tsapkini et al., 2018), contrasting with individuals with svPPA for whom comprehension deficits could limit generalization (Tsapkini et al., 2018; Rogalski et al., 2011), though they are still able to learn (Tsapkini et al., 2018; Hung et al., 2017). Despite strides in identifying a variety of predictors, including structural imaging (brain region volumes and cortical thickness) (de Aguiar et al., 2020a; Nissim et al., 2022), white-matter integrity (Zhao et al., 2021), functional connectivity (Wang D. et al., 2024), baseline language and cognitive abilities (de Aguiar et al., 2020b; McConathey et al., 2017), demographic (e.g., sex; see Licata, Zhao (Licata et al., 2023) and lifestyle factors (e.g., sleep (Herrmann et al., 2022), gaps persist in predicting therapy outcomes and tDCS responsiveness. It is not fully understood which factors predict the effectiveness of tDCS more than others. In this regard, genetic factors also appear to play a significant role.

Individual differences in genetic makeup may play a significant role in shaping the efficacy of neuromodulatory interventions such as tDCS. Genetic factors that influence synaptic plasticity, neurotransmitter function, or vulnerability to neurodegeneration are hypothesized to interact with tDCS-induced neuroplasticity. Although this interaction remains untested in PPA (Wang Z. et al., 2024), related evidence from other neurodegenerative disorders, post-stroke aphasia, and healthy aging provides a compelling theoretical basis for examining genetic predictors in this context.

The brain-derived neurotrophic factor (BDNF) gene plays a central role in regulating synaptic plasticity, neuronal survival, and long-term potentiation and fluctuations in its levels are associated with neuropathological changes observed in neurodegenerative disorders (see Azman and Zakaria, 2022; Ibrahim et al., 2022) for reviews on the molecular mechanism of action). A common single-nucleotide polymorphism, Val66Met (rs6265), is associated with decreased activity-dependent secretion of BDNF (Lamb et al., 2015) which may impair learning and memory; however, this finding is not supported by Bugge Kambestad et al. (2023). Structurally, individuals with the Val66Met gene show smaller hippocampal volume (Molendijk et al., 2012; Hajek et al., 2012; Kambeitz et al., 2012) and cortical thinning in regions relevant to PPA, such as the left frontal and temporal lobes (Yang et al., 2012), potentially impacting their responsiveness to tDCS-based interventions. Functionally, the Val66Met genotype is associated with greater decline in episodic memory and learning impairments (Lim et al., 2021; Cechova et al., 2020; Thomson et al., 2024). In individuals living with post-stroke aphasia, BDNF polymorphism has been linked to aphasia severity and stimulation-induced neuroplasticity (Dresang et al., 2022). In an RCT involving tDCS and language therapy in individuals living with post-stroke aphasia, BDNF val/val carriers showed greater benefit from a-tDCS and aphasia therapy than val/met carriers (Fridriksson et al., 2018). However, this advantage was not observed in a study of aphasia recovery without neuromodulation (de Boer et al., 2017).

The catechol-O-methyltransferase (COMT) gene encodes the COMT enzyme, which plays a crucial role in the degradation of dopamine, particularly in the prefrontal cortex. The Val158Met polymorphism affects enzymatic activity, with Met carriers exhibiting lower COMT activity and thus higher extracellular dopamine levels (Schacht, 2016), which may confer advantages in certain cognitive tasks but can also lead to inefficiencies under high-load conditions. The COMPT effect may be modulated by sex (Elton et al., 2017; Fan et al., 2020). In cognitively unimpaired older adults, COMT Val carriers benefited significantly from tDCS more than those with met/met genes (Hayek et al., 2021).

Finally, the APOE4 allele, particularly in its homozygous form, is a well-established genetic risk factor for Alzheimer's disease (AD), influencing Aβ plaque formation, tau aggregation, and synaptic loss (Sun et al., 2023). It has also been implicated in altered neuroplasticity and cognitive function, even in persons without AD (Corbo et al., 2023; Zimmermann et al., 2019). In the context of PPA, APOE4 is associated with increased β-amyloid deposition and disrupted neural connectivity, contributing to the disease's pathophysiology (Singh et al., 2023). However, its effects are heterogeneous across different phenotypes and subtypes of the disease. Homozygosity is linked to more severe disease progression and potentially poorer treatment response. Moreover, emerging evidence suggests that the APOE4 genotype may influence the effectiveness of tDCS. A recent study by Kang et al. (2021) found that cognitive remediation combined with tDCS slowed decline, with significantly greater benefits observed in APOE4 non-carriers. Understanding the specific pathways affected by APOE4 can guide the development of targeted interventions that may mitigate its adverse effects on PPA progression.

Taken together, BDNF, COMT, and APOE genotypes represent biologically plausible moderators of tDCS efficacy, influencing structural integrity, excitatory-inhibitory balance, and neuroplastic capacity. The present trial is the first to investigate these interactions in PPA and may help identify genetic profiles associated with enhanced responsiveness to neuromodulatory therapies.

The role of biomarkers in relation to tDCS outcomes in individuals living with PPA remains largely unexplored, though several markers may offer insight into disease progression and therapy response. Neurofilament light chain (NfL), a neuronal cytoskeletal protein that plays a crucial role in maintaining the structural integrity of axons, is recognized as a valuable biomarker in the context of neurodegenerative disorders. It is released during neuroaxonal damage, making it a sensitive-but-not-specific biomarker for neurodegenerative processes (Matias-Guiu et al., 2019; Steinacker et al., 2017; Lin et al., 2019). The impact of NfL on therapy response in individuals living with PPA, however, remains unknown.

Additionally, tau is a microtubule-associated protein primarily found in neurons, which, when abnormally phosphorylated or fragmented, can aggregate and contribute to the development of neurodegenerative disorders. It can be an important biomarker in PPAs associated with tauopathies (Shir et al., 2024), and its presence in biological fluids may offer insights into disease progression and treatment efficacy, enhancing clinical utility.

Finally, the impact of amyloid-beta 42 (Aβ42) on language therapy efficacy in PPA is complex, encompassing neurodegeneration and therapeutic strategies. In individuals with lvPPA, the presence of Aβ42, signifying Alzheimer's disease, has been found to correlate with increased cortical atrophy, especially in the left hemisphere, and language deterioration, potentially influencing therapy results (Rogalski et al., 2019). This left hemisphere predominant atrophy correlates with a more significant decline in language performance, potentially reducing the efficacy of language therapy (Rogalski et al., 2019). To our knowledge, there has been no research exploring the interaction between Aβ pathology and neural mechanisms to better understand their combined impact on treatment efficacy in PPA. Together, these biomarkers could eventually inform individualized prognoses or treatment decisions; however, their role in moderating or mediating response to tDCS remains an open and important area for future investigation.

Similar to genetic factors and biomarkers, lifelong language experiences, such as bilingualism, and lifestyle factors, including exercise and sleep habits, have been argued to influence cognitive reserve and neural plasticity, thereby affecting the efficacy of therapy. For example, bilingual individuals with post-stroke aphasia have been shown to benefit more from therapy and have better recovery than their monolingual peers (Masson-Trottier et al., 2022; Lahiri et al., 2021). However, to our knowledge, bilingualism in individuals living with PPA has not been investigated as a predictor of therapy outcomes (Norhan et al., 2023). Exercise has also proven to maintain and potentially improve cognitive functions in aging (Krivanek et al., 2021) and individuals with post-stroke aphasia (Lorenzen and Murray, 2008). Understanding the interplay between these factors and therapy outcomes could provide valuable insights for optimizing treatment approaches and enhancing clinical care for individuals with PPA.

In summary, studies focusing on behavioral treatment of common symptoms in individuals living with PPA, such as anomia and agraphia, have shown promising results in terms of improvements in trained items immediately following intervention, as well as generalization to untrained items and retention of therapy gains weeks after the intervention ends (Robinaugh and Henry, 2022; Taylor-Rubin et al., 2021; Cadorio et al., 2017; Wauters et al., 2023). Neuromodulation methods, especially tDCS, have further enhanced these outcomes, promoting the maintenance and generalization of language behaviors (Cotelli et al., 2020; Coemans et al., 2021; Nissim et al., 2020). However, a Phase II multisite study has yet to be performed to evaluate large-scale efficacy and further assess safety, examining the benefits of tDCS in individuals with PPA. Furthermore, predicting therapy outcomes in PPA remains complex, influenced by factors such as PPA variant, neuroanatomical and neurofunctional factors, and baseline cognitive abilities. Significant gaps remain in understanding the roles of genetic factors, biomarkers, and lifelong language experiences in shaping therapeutic outcomes, particularly with tDCS. Addressing these gaps is important for optimizing treatment approaches and improving clinical care and neuromodulation outcomes for individuals with PPA.

Despite promising findings from the previous single-site Phase II study using tDCS paired with similar language therapy in PPA (Tsapkini et al., 2018) and other studies from members of our group (Roncero et al., 2017, 2019; Hung et al., 2017; Nissim et al., 2022; McConathey et al., 2017), no multisite Phase II trial has systematically evaluated its efficacy, which is a helpful step to facilitate transfer of tDCS into the clinical environment. This gap is significant given the encouraging learning and generalization effects observed in lvPPA and nfvPPA populations. The present study is designed to address this gap by evaluating the impact of combined tDCS and behavioral therapy on naming and spelling in a well-powered multisite randomized controlled trial (RCT), and by identifying clinical, neural, and genetic predictors of treatment response. A crossover RCT enables each participant to serve as their own control, increasing power and reducing inter-individual variability.

We will include 120 participants with PPA who are between 50 and 80 years of age, right-handed, proficient in English, and have a minimum of a high school education. Participants must be diagnosed from a PPA and Alzheimer's disease (AD) clinics at the University of Pennsylvania, Johns Hopkins University, the University of Toronto, or other specialized referral centers in the US and Canada, using current consensus criteria (Gorno-Tempini et al., 2011). To corroborate the diagnosis, AD biomarkers (CSF or PET amyloid β I-42) will be examined and we will use our variant classification tests developed, including a spelling task and a speech production task (Cookie Theft picture description task (Goodglass and Kaplan, 1983), which can accurately discriminate PPA participants with over 80% accuracy (Themistocleous et al., 2021; Neophytou et al., 2019).

Participants are asked to maintain their daily habits as consistently as possible throughout the trial. Therefore, individuals already receiving speech-language therapy before enrolling in the project may continue to see their therapist. However, participants who were not already seeing a therapist are requested to wait until the end of the trial before starting any new therapy sessions.

We will exclude individuals who have any of the following: a history of neurological disease, including vascular dementia, large vessel stroke, or attentional deficit; uncorrected hearing loss or visual acuity loss; a history of major pre-morbid psychiatric disorders; left-handedness; or advanced dementia impairments [Mini-Mental State Examination (MMSE) (Folstein et al., 1975) < 15] or severe language impairments preventing the participation in assessments and therapy.

In this trial, the overarching aim of the intervention is to enhance communication effectiveness in individuals living with PPA by improving their spoken naming and spelling of nouns and verbs. NaSp was developed based on our previous work (Tsapkini et al., 2014, 2018), and improved based on additional evidence from the literature to enhance spoken naming specifically (Rapp and Glucroft, 2009; Tsapkini and Hillis, 2013; Beeson et al., 2002). While NaSp includes elements from these earlier therapies, it also incorporates modifications that target spoken naming abilities—an aspect not effectively addressed by the therapy used in our prior studies.

Using the Rehabilitation Treatment Specification System (Hart et al., 2019) as a framework, NaSp can be described as a multi-stage intervention that is decomposable into therapy targets, each of which is linked to specific active ingredients. Specific Skills and Habits targets include improving spoken naming and spelling accuracy of trained and untrained nouns and verbs, increasing metacognitive skills to compensate for anomia, and enhancing language and cognitive skills to support communication. An Organ Functions target is to improve the efficiency of the language network in the brain in the presence of neurodegeneration. Finally, Representation targets are to increase confidence in communication and maximize engagement and collaboration. While breaking NaSp down into its specific active ingredients provides a detailed therapeutic perspective, it goes beyond the scope of this paper and will be addressed in a subsequent publication.

During NaSp therapy sessions, each trial consists of two phases –spoken naming and spelling. Figure 2 illustrates the flow of NaSp therapy. Each trial begins with the spoken naming phase, in which individuals with PPA are presented with a picture and asked to name it orally. If successful, they proceed to write the word. If unsuccessful, they receive prompts to generate three semantic cues to aid in naming, followed by a phonemic cue if needed. During the cueing, participants are encouraged to name the picture as early as they can (i.e., they do not need to wait for the three semantic cues). If the participant can still not name the picture, the spoken word is provided, and they are asked to repeat it three times before moving on to spelling. In the spelling phase, participants are asked to write the word corresponding to the picture. Successful attempts are reinforced by confirming the provided answer is correct, while unsuccessful attempts prompt PGC training–also known as sound-to-letter correspondence–for the target word being trained (Tsapkini and Hillis, 2013). Following the PGC training, participants must attempt to write the word corresponding to the picture a second time, with reinforcement provided upon successful completion. If unsuccessful, they are guided through the writing process: the speech-language therapist (or a trained research assistant), hereafter referred to as the therapist, shows the word again and provides PGC training while the participant copies it. In cases where participants improve substantially within the first sessions of intervention, the therapist encourages the participant to form sentences using the target words or engage in personally relevant conversations about the target word to ensure all participants receive at least 45 min of therapy.

As shown in Table 1, Figure 1, a therapist will administer therapy 5 days per week for three consecutive weeks. Therapy will last between 45 and 90 min, including the concomitant stimulation during the first 20 min. The exact duration of therapy will vary depending on the participant's tolerance for effort and speed of execution, and breaks will be taken if needed. Therapy will be conducted individually, in person, in a quiet room.

During each therapy session, the 20 words on the trained list (described in the next section) will be used for that period. To ensure therapy validity across different sites, a NaSp app was developed by the first author (MMT) in R using Shiny App. The app guides the therapist in following the NaSp flow according to the participant's answers. It allows the therapist to track the participant's spoken and written naming performance, as well as the effectiveness of cues when used. A comprehensive NaSp tutorial was also developed to help train therapists at the different sites.

The administration of tDCS involves using the Soterix 1x1 platform and 5 cm × 5 cm electrodes. The anode is positioned over the left frontal lobe at F7 according to the 10-20 electrode placement system, while the cathode is placed extracephalically over the right cheek. This unihemispheric tDCS montage aims to achieve more targeted stimulation, aligning with predictions from prior modeling studies (Moliadze et al., 2010; Wagner et al., 2007; Bhattacharjee et al., 2019). While the F7 position provides a standardized landmark for left IFG stimulation, it is important to acknowledge that individual anatomical differences such as cortical folding, sulcal depth, and skull thickness can influence the intensity and distribution of the electric field delivered to the target region (Laakso et al., 2015; Saturnino et al., 2019). However, modeling data in neurodegenerative populations suggests that such variability may not substantially compromise targeting, despite atrophy-related anatomical changes in individuals with PPA, consistent current flow patterns to target regions remained consistent across variants and individuals (Unal et al., 2020). While individual electric field modeling can improve focality, it remains resource-intensive and challenging to implement in large clinical trials. Moreover, in the context of neurodegenerative conditions like PPA, where functional reorganization and loss of network hubs may occur, increased focality may not always be advantageous (Mandelli et al., 2018; Agosta et al., 2014; Tao et al., 2020). In fact, when comparing conventional and high-definition tDCS, conventional tDCS seems to yield stronger effects, and it could be the lack of focality that make it very efficacious as it is discussed in Unal et al. (2020). These findings support the robustness of this tDCS montage in achieving reliable stimulation even in the presence of disease-related anatomical variability. See Figure 3 for the simulation of current propagation with the montage used in this study.

To minimize skin-electrode reactions, a non-metallic, conductive rubber electrode with a saline-soaked sponge is used to cover the left inferior frontal gyrus (IFG). Each tDCS session delivers 2 mA of current (with an estimated current density of 0.08 mA/cm²) for 20 min during the a-tDCS period, concurrent with the naming therapy, and is followed by at least 25 min of therapy alone or the time required to complete the trained list.

During sham stimulation (s-tDCS), a brief 30-s ramp-up and ramp-down of current mimic the sensation of active tDCS. This method has effectively blinded participants to the stimulation condition in prior studies (Palm et al., 2013). In earlier studies conducted by our team, including randomized, sham-controlled trials of tDCS in individuals with PPA, blinding procedures were shown to be effective. Specifically, post-trial assessments of participant and care partner guesses regarding stimulation condition (active vs. sham) revealed accuracy rates at chance (Tsapkini et al., 2018) or were unable to identify a difference (Roncero et al., 2019), suggesting successful blinding in this population.

To monitor adverse effects, each participant is asked to rate their pain level once at the end of each stimulation session using the Wong-Baker FACES Pain Rating Scales (http://www.WongBakerFACES.org) (Wong-Baker FACES Foundation, 2022). The protocol requires each participant to undergo 15 consecutive weekday sessions.

After completing the trial, at T7, participants will also be asked to indicate during which period they believe they received active or sham stimulation (using a dichotomous response format). This protocol ensures comprehensive data collection while maintaining participant safety and minimizing potential sources of bias.

The primary outcome measure will be the change in performance between time points, as measured by phonemic accuracy (during spoken naming) and letter accuracy (during spelling) of trained items. Phonemic and letter accuracy will be calculated on a scale of 0–100%, with a higher number reflecting higher accuracy, using the Phonology analysis and Spelling analysis tools in Open Brain AI (Themistocleous, 2023).

At each assessment time point, participants undergo a comprehensive evaluation that includes the NaSp spoken naming and spelling task, which serves as the primary outcome measure. The task consists of 80 stimuli corresponding to the Level 1 stimuli previously outlined in the Therapy Stimuli section (along with 80 items included in Level 2 for identified participants). Accuracy on trained items is then computed separately.

The assessor presents all the words in the list. For each noun, participants attempt to name the item orally and then write their answer. No feedback is given, as some items serve as generalization items for either P1 or P2. After assessing the nouns, the same procedure is applied to the verbs.

All spoken and written responses are recorded. Spoken responses will be transcribed manually using the International Phonetic Alphabet, and written responses are captured using a smart pen (Symphony Smartpen by Livescribe, n.d.) to ensure accurate data collection. Both the spoken and written responses will be typed into a scoring document.

The secondary outcome measures will be the changes in performance between time points, specifically phonemic accuracy (during spoken naming) and letter accuracy (during spelling) of untrained items. Similar to the primary outcome measure, phonemic and letter accuracy will be calculated on a scale of 0–100%, with higher numbers indicating greater accuracy, using the Phonology analysis and Spelling analysis tools in Open Brain AI (Themistocleous, 2023). The performance on the NaSp spoken naming and spelling task will be used to compute the accuracy of the untrained items.

Additionally, resting-state functional MRI (rs-fMRI) will be used to measure the activity of various brain regions during resting or task-negative conditions. This approach will allow for the evaluation of functional interactions between brain regions. Further details regarding the neuroimaging methods are provided below.

MRI protocols for structural and resting-state functional MRI will follow those established by the Alzheimer's Disease Neuroimaging Initiative (ADNI). This large, multi-center, and multi-manufacturer study offers standardized protocol implementations that can be harmonized across different MRI devices, ensuring consistency and comparability of imaging data across various research sites. Each participating site will adapt the ADNI-3 protocol to their specific scanner hardware and software while maintaining core acquisition parameters to ensure harmonization. Structural and functional connectivity will be measured using the following methods based on the ADNI-3 protocol (representative parameters):

Saliva samples will be collected for processing to extract DNA. Saliva samples will be stored at room temperature and processed in batches.

Blood samples will be collected and processed to extract relevant biomarkers.