What is the evidence for interoceptive and allostatic dysfunction in frontotemporal dementia?

I appreciate the authors careful revisions and responses to my prior comments. I feel the manuscript and analysis plan are much clearer. In particular, I think the authors are much clearer about their assumptions (and the limitations thereof), which will better help position their work in the broader literature. I have just a few minor pieces of feedback, but am happy to see this paper move foreward.

1. I'm still a bit stuck on how interoception and allostasis are being differentiated. It may be worth clarifying that allostasis incorporates sense data from all sensory modalities, not only interoceptive. As it's written, these two words still seem synonymous. 

2. Related to #1, it's not immediately clear based on the authors definitions and search criterion how an index like "heart rate" or "immune markers" are related to interoception per se. To better marry the discussion on interoception and allostasis it may be worth just saying explicitely that these indices represent sensory events that are encoded by interoceptive sensory cells in the periphery/brain AND can be dysregulated in the case of allostatic load.  Right now the latter point is clear, but the prior isn't clearly demarcated. 

3. "We will also exclude proprioception and vestibular function from our definition of interoception, since they represent the position and movement of the body in space rather than its physiological condition, and are not linked to homeostatic/allostatic regulation but to action" - just a note that folks like Barrett and colleagues have argued that visceromotor control ("homeostatic/allostatic regulation") is a kind of action, just like skeletomotor control so this distinction may not be clear. 

While dementia has historically been—at least heuristically—considered a neurocognitive disease, or a disease solely of the brain, the authors do an excellent job outlining how various forms of dementia are associated with functional decline in several physiological systems (e.g., inflammation, metabolism) and implicates changes to brain regions supporting physiological representation and regulation. I agree with the authors that “the current literature fails to provide a thorough investigation that combines these aspects cohesively.” To this end, I feel strongly that the authors’ proposed project will be an extremely valuable contribution to the existing literature.

My main suggestions center on how the authors are characterizing interoception and allostasis/allostatic load.  These larger comments may shape how the authors choose to move forward in section 2.1.2.

1. Interocepton and allostasis:

1.1  Given the review questions, it may be useful to disaggregate physiological and survey/task-based notions of interoception. In our own work, we’ve begun referencing “the interoceptive system” (encompassing sensory cells, transmission pathways, and brain systems) and “interoceptive traits” (encompassing individual differences in self-reported or task-based interoceptive ability). We’ve argued that the interoceptive system gives rise to these traits (akin to how the visual system gives rise to individual differences in visual acuity) but is not necessarily synonymous to them. Some set of similar descriptive labels may help you disaggregate and clearly describe findings later on given that you are hoping to annotate biological and behavioral markers of interoceptive and allostatic dysfunction.

1.2  As the authors are aware, there is still wide disagreement on how broad or narrow conceptualizations of “interoception” should be. The authors include a very broad definition “including the skin and skeletal muscle.” They also highlight current work suggesting that efferent and afferent pathways of the allostatic-interoceptive system may be more unified that historically considered (you may also be interested in an awesome paper by Clare Shaffer on sensory coding in the Vagus which also argues for greater afferent and efferent integration!). I’m extremely sympathetic to these ideas. However, there are a few points where I think some more specificity or clarification would be useful:

• The authors describe a definition of interoception based on its “physiological, homeostatic role” and on the “functional and neuroanatomical organization” of pathways converging in the AIN (see point below). Without further explanation, this definition of interoception is very unclear and doesn’t, for example, explain why olfaction, proprioception, and vestibular function–but not taste—are omitted in your search. Without further explanation—these choices seem random. I can understand (and in fact, applaud) the motivation to be expansive in your definition of interoception when trying to begin outlining how dementia may impact the body-brain system. Future work can add additional specificity. However, I think it may be more defensible to simply say this explicitly (acknowledging that broad definitions like this have received criticisms) instead of trying to bend over backwards to define interoception in a way that allows you to look across many physiological systems in the literature.
• Related to the point above, the authors do not discuss to what extent they consider somatosensory sensation part of “interoception” when they mention skeletal muscle. It’s not clear whether they are saying that interoception ought to include somatosensory signals from the bones or muscles, or if they are trying to include the role of more canonical interoceptive channels in processes like bone metabolism. While there is certainly evidence for interoceptive and somatosensory integration (e.g., in the brainstem), there do seem to be meaningful differences in the computations and thus, qualia of sensation afforded by these systems. Also, to the extent they implicate different swaths of brain tissue this may be important for discussions on dementia.
  
  
• By my understanding – “allostasis” reflects both the representation and regulation of the body (i.e., the sensory and motor arms). To this end, allostasis involves interoception (the sensory arm) and interoception contributes to allostasis. The allostatic interoceptive network, as described by Barrett and colleagues thus encompasses regions involved in interoception and visceromotor control. Throughout the paper, the terms “interoception” and “allostasis” are used in ways that often seem interchangeable. I would either adjust or be clearer about your assumptions. This will also help when describing results down the line. Relatedly, I wouldn’t necessarily say that interoceptive information “converges within the allostatic-interoceptive network.” The network is large and so this sentence is a bit confusing. Maybe, “converge within brainstem nuclei of the allostatic-interoceptive network” or something like that. This may also inform 2.1.2 number (2).
  • Relatedly, I was confused by the phrasing , “Interoceptive-allostatic overload” (which was never defined).

2.        Allostatic Load: I Would take more time considering exactly what you mean by “allostatic load.” This construct is infamously difficult to describe and operationalize. In particular, it’s probably worth clarifying that you cannot really have any single marker of “allostatic load” – given that “allostasis” within the predictive coding framework is really about coordination (e.g., prioritization of resources) across body systems and “allostatic” load is really about imbalance or poor coordination. I’m wondering if there is a more concise way of discussing this construct in the context of your proposed methods. Specifically, I’m thinking about McEwan & Seeman’s (2006) paper “Protective and Damaging Effects of Mediators…” where they describe allostatic load in terms of primary, secondary, and tertiary outcomes.  This framework was used successfully in a recent paper by Buller-Peralta et al (2024). This may or may not work given your plans but—may be worth exploring as a way to further organize papers?

All in all I am excited by the prospect of this work and, as I said before, believe it will contribute to our understanding of how body-brain pathways are disrupted in, and contribute to dementia pathology. 

Thank you for the opportunity to review your scoping review protocol and for your patience with my return of this review.  Given that this is a proposal for a scoping review, many of the review suggestions/criteria suggested by PCI RR do not really apply here.  Below, I provide a few suggestions that I think are worth considering while the authorship team carries out the scoping review.

This is an incredibly important topic. There is reason to believe that interoceptive changes and challenges in neurodegenerative diseases may be driving some of the other challenging symptoms (like changes to cognition) and the distortion of sort of meta-monitoring that is required for interoceptive sensing is thought to be correlated with caregiver burden.  On top of that, we have some new evidence that in very early stages of neurodegenerative disease(s) in a monkey model, interoception seems to be on the leading edge of symptomatology (Baxter et al 2024, SfN abstract). FTD in particular – as noted by the authorship team – targets brain regions that are thought to be critical for interoceptive processing. 

The protocol for the review is sounds and clearly follows the PRISMA organization.  The only other component that might be worth considering is looking at who/what papers are citing the papers that you turn up in your lit review.  This is kind of the opposite of searching the papers reference sections and I think this approach is not typically included as a method in these sorts of reviews – but we have found that it sometimes turns up papers of importance.

The conceptualization of interoception and allostasis as detailed in the document are fine but I would propose one additional distinction (made to some degree in the background) which I think may be important for the review and making sense of the data – the distinction between the generation/processing of interoceptive signals and the awareness of those signals (as would be recorded, for example, in a heartbeat detection task).  The authors seem to be lumping these two types of data together – or if they are not, it is not totally clear. Some of this lack of clarity actually has to do with definitions.  As they point out Sherrington coined the term interceptors to refer to the information processing.  The Craig definition which has become common parlance leans on the conscious representation of interoceptive information.  I would separate these two types of studies as one might expect that self-reports of physiology deteriorate or are impacted on a different time/disease course than the actual physiology itself. 

It is not clear to me why taste is being included and olfaction is not.  I probably would not include either as they seem to be exteroceptive senses rather than interoceptive senses, but my suggestion would to be clear about why particular sensory streams are either included or excluded. 

With regards to demographics variables:

It would be very interesting if it was possible to comment on gender differences in these effects.  I have no idea of the papers that the group will turn up will separate genders in their analyses to allow for such a comment, so such a comment may be impossible.  

Similarly, if it is possible to include information about the timing of interoceptive assessment vis a vis the timing of diagnosis or stage of diseases progression, I think that could be very helpful to paint a picture of when interoceptive challenges occur in the time course of disease.