Regional patterns of human cortex development correlate with underlying neurobiology

Human brain morphology undergoes complex changes over the lifespan. Despite recent progress in tracking brain development via normative models, current knowledge of underlying biological mechanisms is highly limited. We demonstrate that human cortical thickness development and aging trajectories unfold along patterns of molecular and cellular brain organization, traceable from population-level to individual developmental trajectories. During childhood and adolescence, cortex-wide spatial distributions of dopaminergic receptors, inhibitory neurons, glial cell populations, and brain-metabolic features explain up to 50% of variance associated with a lifespan model of regional cortical thickness trajectories. In contrast, modeled cortical thickness change patterns during adulthood are best explained by cholinergic and glutamatergic neurotransmitter receptor and transporter distributions. These relationships are supported by developmental gene expression trajectories and translate to individual longitudinal data from over 8,000 adolescents, explaining up to 59% of developmental change at cohort- and 18% at single-subject level. Integrating neurobiological brain atlases with normative modeling and population neuroimaging provides a biologically meaningful path to understand brain development and aging in living humans.

Introduction.The manuscript does not provide sufficient detail on the potential mechanisms by which these molecular and cellular processes influence the development of cortical thickness.A further description would be beneficial.
The brainchart data are from a previously published study (Rutherford et al.).However, from the age distribution shown in Fig. S5, there are very different numbers of subjects at different ages.In particular, there are very few subjects before age 10 and around 40.Is the brainchart data affected by these confounding variables?Lines 1-10, Page 6.The authors show the spatial associations between cross-sectional thickness and neurobiological features and describe diverse colocalization trajectories.However, the results are not clearly described.The authors found a general pattern of strongest changes from childhood to young adulthood (up to approximately 30 years) as well as in late adulthood (from 60 years onwards; Fig. S7).A possible reason for this result is that the neurobiological maps are from adult brains.Figure S7 should be moved to the main text.The authors stated that sex did not relevantly influence the trajectories (Fig. 8).Did they perform statistical comparisons for sex differences?
The authors showed the spatial associations between neurobiological markers and changes in cortical morphology.Several studies have demonstrated a network mechanism underlying the maturation of cortical morphology (e.g., PMID: 27457931;PMID: 38278807).I would suggest adding a discussion on this topic.
The curves presented in Fig. 2 appear to differ from those in Fig. 3A.It is unclear which result should be referenced to understand which molecular and cellular markers more significantly constrain cortical thickness development at specific ages.Please clarify.
Line 21, Page 2. Introduction.The authors describe that "neurodevelopmental disorders are associated with both deviations in brain structure and dysfunction of several neurotransmitter systems, but suffer from a lack of reproducible biomarkers and little clinical translation of neuroimaging research.".Many previous studies have reported that neurodevelopmental disorders such as autism and depression are associated with deviations not only in brain structure but also in functional connectivity.Moreover, the reproducibility of results has been well studied using multisite neuroimaging data.The authors need to rephrase their claims.In addition, this paragraph describes the importance of neurotransmitters and the normative model in understanding disease or atypical development.However, this is not reported in the results section, which is confusing.Line 5, Page 3. Introduction.The authors described that "multimodal neuroimaging-based spatial association approaches can provide a window into specific biological mechanisms, but until now were limited to postmortem data and the cellular level." In fact, there are many multimodal neuroimaging studies showing spatial associations between brain signatures such as brain connectivity and metabolism (PMID: 23319644; PMID: 33443160; PMID: 30741935).
Line 30, Page 3. 21 neurobiological maps should be presented in the main text, given their importance for the association analysis.Results: Ni8 was not mentioned.
Line 18, Page 6.The authors showed that molecular-or cellular-level markers explained up to 54% of the spatial variance.How about the combined molecular and cellular-level markers?How about the sex differences?Line 8, Page 8.In the main text, the description of Fig. 3A states, "We found that the nine FDRsignificant molecular and cellular markers jointly explained 58% of CT change patterns from 5 to 30 years."However, Fig. 3A is labeled as "neurobiology markers significant in univariate analyses."This raises the question of whether the analysis is multivariate or univariate.Please clarify.
In Fig. 6, the meaning of the colored overlay is unclear.What do different shades of the same color represent (for example, in the context of synaptic pruning) (Huttenlocher & Dabholkar, 1997)?
The discussion section lacks an in-depth exploration of the physiological significance reflected by the constraint of relevant neurobiological markers on cortical maturation at specific developmental stages.This should be a central concern of the paper and needs further elaboration.
Reviewer #2 (Remarks to the Author): The study integrates numerous "multilevel" cortical atlases, normative trajectories of MRI based cortical thickness and partly separate population-based longitudinal youth samples in an attempt to improve our understanding of the neurobiology underlaying cortical thinning across the life span.I find the study comprehensive and highly relevant for a broad field within MRI based neuroimaging i.e. all studies assessing cortical macrostructure.
My comments are addressed below in chronological order and I sincerely hope they can be of some help.As this is a multidisciplinary study, I would like to state that I am a neuroscientist and refer to other reviewers to better scrutinize topics outside of neuroimaging.Also, please do not hesitate to correct me whenever I have misunderstood something! 1. Introduction: I believe reference nr. 4 regarding the link between MRI based cortical development and cognitive development from the year 2000 is dated and could be updated.2. Introduction: I find the use of the terms "development" vs "change" somewhat unclear, particularly when attempting to grasp the underlying logic of the paper i.e: "Assuming that CT changes over the lifespan are shaped by activity, development, or degeneration of cell populations and molecular processes, it is to be hypothesized that the spatiotemporal patterns of CT development colocalize with nonpathological adult spatial distributions of the respective neurobiological marker.Increased colocalization of CT changes with an individual marker at a given developmental period would then support a role of the associated cell population or process in respective CT changes.»For the most part it appears that development is not simply "change over time", which then could pertain to the full lifespan but instead more closely relates to "maturational" processes found in youth only, is that correct?And that for the adult part of the lifespan the word "change" is often used to capture a broader set of processes?I think different disciplines use the word "development" quite differently, so maybe either define in a word or two or be very concise about usage.3. Introduction: The example paragraph from comment 2 holds vital logic for the paper but was (to me) a heavy read, where I had to dissect every sentence.Is there a way to, when presenting this very central concept for the first time, do it "better"/simpler?-I think this text captures why it would make sense to map cortical thinning in youth onto adult biological architecture, but I could not fully grasp why that is not problematic.Particularly as the introduction previously states that different mechanisms might underlay youth based-and adult thinning.
-The reader could also be guided better through this very first introduction of increased co-consistent in the terminology and only use development and change when talking about longitudinal data like ABCD/Imagen.12. Discussion: It was slightly unclear to me whether your findings match the previous main candidate mechanisms for cortical development.To my understanding re-organization of dendritic arbour, cortical myelinization, and even synaptic pruning (although controversial as MRI studies cannot be solely explained by small-scale changes at the synapse level) are candidates.Your discussion instead highlight microglia and thus immune responses and dopamine receptor activity is that correct?When I read the discussion, it currently appears as if all your results fit previous literature with no discrepancies discussed.Also, could the relation to the dopamine system be considered controversial just in regard to the mm scale of MRI? 13.Discussion: It is stated that "As promising candidates for clinical translation, we identify the dopaminergic system and microglial cell populations for early development".Would one not have to apply the normative models to clinical populations like in the Rutherford paper, in order to probe clinical relevance?It is unclear to me why non-typical age trajectories should relate to mental health any more than to any other variable that shows stronger associations to imaging and cortical thickness specifically.Like a previous comment, I find the link to clinical utility inconsistent, and in my opinion not necessary for the relevance of the paper.14.Methods: It is stated that "Given that the main goal of the current analysis was to establish the feasibility of capturing associations between CT development and multilevel neurobiological markers on the individual level, clarification of the sources of .. site-effects will be a task for future investigations.When using multi-site initiatives like ABCD (21 sites, 29 scanners), harmonizing or somehow minimizing the effects of scanner is basic convention.It is not clear why the authors find this a task for the future as the large effects of scanner is already widely documented also specifically within ABCD.The normative models used within the paper attempt to minimize effects of scanner, should the same not pertain to the target longitudinal data?Please explain why scanner bias is not relevant for your analyses or add it as a limitation.15.Methods It is stated that site effects were estimated in «healthy subsamples» of both dataset's baseline data (n = 20 per site, 50% female) how was healthy defined?Reviewer #3 (Remarks to the Author): I previously reviewed this paper for Nature.I have not retained a copy of the version submitted to Nature, and the authors have not provided a rebuttal letter detailing their response to reviewer comments on that submission, so it is difficult to be sure if/how this version has been revised compared to the version previously reviewed for Nature.I think some of my points have been minimally addressed in Discussion, however there don't appear to have been many, if any, substantive changes.Most of my prior comments therefore still stand, as copied below: "This paper reports the results of an extensive set of analyses, using multiple large prior datasets, conducted by an expert and experienced team using careful and generally well-described methods.
Overall the text and figures are composed to a very high standard.
The study investigates spatial colocation of cortical thickness (CT) with multiple molecular imaging maps (referred to as nuclear imaging) and maps of neurotransmitter density and cellular density derived from postmortem data.The basic concept of colocalising MRI phenotypes with genomic, molecular and cellular reference atlases in an effort to understand more about the neurobiological substrates of MRI phenotypes is not novel.This strategy has been widely used in recent years and it has generated interesting results, as in this paper, but it also has well-known limitations.It is clear that the authors are aware of all this.However, I am not sure how decisively this work advances the field beyond what has already been done, or tackles some of the fundamental limitations of spatial colocation analysis for mechanistic interpretation of MRI phenotypes.