Anorexia Nervosa in the Eyes of Neuroimaging: A Meta-Analysis of fMRI and MRI Studies

See the above slides for further information and statistics on anorexia nervosa.

Background

Anorexia nervosa (AN) is considered one of the deadliest psychiatric disorders impacting the global population, where 5-20% of all diagnosed individuals will succumb to complications related to this condition. Not only is AN incredibly challenging to identify, individuals diagnosed with AN face staringly high relapse rates and a lack of sufficient, modern treatments. In order for patients suffering with AN to lead successful, fulfilling lives, it is critical that we accept the brain’s physical role in the development and behaviors of AN.

Purpose

To determine a set of brain regions that have significant structural or functional differences in patients with AN compared to healthy controls (HC), by investigating MRI scans of white matter (WM) and grey matter (GM) volume, as well as resting-state connectivity via fMRI (fMRI).

Approach

Perform a coordinate-based meta-analysis by screening for studies that contain one/more of the three modalities (fMRI, GM, WM) and then conducting analyses on the included studies to isolate abnormal clusters of volume or activity in AN.

Impact

This project aims to establish potential biomarkers or regions of interest (ROIs) that can improve both the diagnosis and treatment of future patients with AN. By determining select regions that consistently show abnormalities in AN across numerous studies, we can better understand the brain structures involved in the disorder and the functions they control. Thus, it can lead to more accurate diagnostic criteria and directed medications or therapies.

Selection

Article Retrieval

Locate all existing, open-access, meta-analyses studying any biomarker in AN and collect their references.

Obtain primary studies researching the biomarkers of AN from across seven scientific databases, through filters and search strings.

Title and Abstract Screening

Included studies are screened based on strict criteria established according to which populations, study designs and methodologies best meet the goal of the analysis.

Phase I:

Remove studies that are not primary research articles, are inaccessible or unrelated to biomarkers in AN.

Phase II:

Remove studies that do not include HC and individuals exclusively diagnosed with untreated AN.

Phase III:

Remove studies that do not include any of the three modalities in whole-brain scans.

Full-text Screening

Verify that selected studies comply with all of the inclusion criteria and contain peak coordinates to be extracted in the subsequent data analysis.

This reduced 2112 studies to 33 articles (AN=950, HC=1058).

Analyses

This project uses the GingerALE application from BrainMap to perform Activation Likelihood Estimations (ALEs). An ALE analysis identifies significant clusters based on multiple studies converging in the same brain region.

ALE Analysis

Phase I:

Conduct single dataset ALE for each modality and their subtypes (direction of change).

Overlay cluster maps in a medical imaging viewer to determine regions of multimodal overlap.

Phase II:

Conduct contrast ALE on modality and subtype pairs, which indicates whether there are clusters where one modality/subtype is stronger than another, and gives a precise point of conjunction.

Network Analysis

Overlay cluster maps from all previous analyses on the Yeo 7-Network Atlas, to locate which of the seven key neural networks each abnormal cluster belongs to.

Calculate the total clusters and volume corresponding to each network.

Voxel-wise Correlation

Make overlap masks between modality pairs in Python to compute Pearson correlations of overlapping voxels’ ALE intensity values.

Results

Single Modality

AN was found to be primarily associated with resting-state activation, grey matter volume decrease and white matter volume decrease. The ALE analysis identified 6 clusters of fMRI activation, 3 clusters of GM reduction and 7 clusters of WM reduction. Note that clusters were observed in the opposite subtype, yet were less abundant and convergent.

Most Significant Clusters

fMRI activation in the left Superior Parietal Lobule (Z=4.22)

GM decrease spanning the left Precuneus and Superior Parietal Lobule (Z=4.56)

WM decrease in the left Superior Temporal Gyrus (Z=4.16)

Multimodal Overlap

When single modality cluster maps were overlaid on the same brain image, five key approximated regions showed multimodal convergence:

Left Superior-Medial Frontal Gyrus: fMRI activation, GM decrease, WM decrease

Left Superior Parietal Lobule: fMRI activation, GM decrease

Right Precuneus: fMRI activation, GM decrease

Right Insula: fMRI activation, WM decrease

Left Middle-Frontal Gyrus: fMRI activation, WM decrease

Conjunction Points

The conjunction ALE analysis successfully assigned peak coordinates and ALE values to each of the multimodal clusters described above. This allowed for a precise anatomical label for each region, rather than an estimated point of overlap according to its location on the Montreal Neurological Institute (MNI) brain template atlas. With this data, the highest multimodal convergence was found within the left Precuneus (ALE=0.0092), where both fMRI activation and GM decrease were concurrent. It also determined that patients with AN experience simultaneous fMRI activation and deactivation at a point in the Central Posterior Cingulate Gyrus (ALE=0.0015), a result unique to the resting-state modality.

Contrasts

Out of the three pairwise contrasts, only one comparison showed significant difference in modality strength. GM decrease was stronger than fMRI activation at six distinct clusters. Of these peak clusters, the greatest difference in change between the two modalities was observed in the right Precuneus (Z=2.01, P=0.022) and the left Inferior Parietal Lobule (Z=2.01, P=0.022).

Network Analysis

Clusters of abnormality within AN could be assigned to all seven networks depicted by the Yeo 7-Network Atlas. A vast majority of the clusters fell in the Default Mode Network (DMN), with a total of twelve peak regions, whereas the cluster count of the other networks did not exceed four. Consequently, the DMN had the most affected volume, where a total of 85.196cm³ of brain tissue experienced structural or functional changes in AN. These findings were acquired from the combined single modality and multimodal conjunction clusters, and did not include the regions where the strength of GM decrease was greater than the fMRI activation.

Voxel-wise Correlation

The weak negative voxel-wise correlations within the fMRI activation-GM decrease overlap (N=639 voxels, r=-0.21) and the fMRI-activation-WM decrease overlap (N=935 voxels, r=-0,18) indicate that although these modalities converge spatially, their voxelwise ALE intensity patterns show minimal and slightly inverse alignment. A zero voxel-wise correlation within the GM decrease-WM decrease overlap (N=268 voxels) suggests an unrelated pattern of convergence strength.

Discussion

To the best of my knowledge, this is the first ALE-based tri-modal meta-analysis of whole-brain abnormalities within AN which includes further convergence, contrast, network, and voxel-wise analyses. The findings illustrate that AN is characterized by structural atrophy and connective dysfunction within patients’ brains, where three distinct modalities (fMRI, GM, WM) appear to play important roles in its pathology. Specifically, when compared to HC, regions of overactive resting-state connectivity and reduced grey and white matter volumes are predominant in AN. While the clusters of fMRI activation and WM volume decrease are more abundant and widespread, their convergence is significantly lower than clusters of GM reduction. Moreover, the contrast analysis reveals that GM loss is the strongest modality, suggesting that AN is driven by select clusters of GM volume decrease.

AN is found to be associated with clusters of multimodal change at five key areas, where only the left Superior-Medial Frontal Gyrus (mSFG) experiences tri-modal abnormalities. Inverse voxel-wise correlations between fMRI-GM and fMRI-WM pairs shows a tendency for higher ALE values in one modality to coincide with lower values in the other, despite sharing the same location. The mSFG may contribute greatly to the processes impaired in individuals with AN, due to its involvement in self-referentiation. This region, amongst most of the identified clusters, belongs to the Default Mode Network (DMN). Not only is the DMN hyper-connected in several psychiatric conditions, existing medications successfully modulate its functions. This opens up possibilities for more nuanced behavioral diagnoses and potential treatments for AN.

Improvements

While this research adhered strictly to the PRISMA 2020 guidelines, it could be refined by further studying functional connectivity as patients are engaged in various tasks, encompassing more studies, including different meta-analytic programs, and conducting additional statistical tests and analyses.

Next Steps

An immediate expansion on this work would investigate factors influencing convergence, additional modalities, comorbid or treated AN, and how the pathology of AN compares to other disorders.

Future Research

Long-term research beyond this project will aim to uncover the cellular and molecular mechanisms behind AN in order to determine viable treatments for this debilitating psychiatric condition.

Although I was unable to find a mentor for this project, a few professors were able to meet with me to discuss the early stages of my research and the ideas that I wanted to explore. Special thanks to Dr. Juan Li from the Ottawa Hospital Research Institute for her resources and assistance in starting the meta-analysis procedure, to Dr. Zhi Wei Zeng from the University of Toronto for his knowledge on running biomolecular simulations, and to Dr. Ian Weaver from Dalhousie University for his insight into the field of neuroscience and advice on choosing a research topic.

Lastly, I am forever grateful to my family for always supporting my research projects and for encouraging me to participate in my first ever regional science fair two years ago, and to the entire team at the Halifax Sci-Tech Expo (HSTE) for being so immensely helpful in preparing for the CWSF.

Key Resources

[1] BioImage Suite MNITAL. (n.d.). Bioimagesuiteweb.github.io. https://bioimagesuiteweb.github.io/webapp/mni2tal.html

[2] BrainLine. (2011, April 22). Brain Imaging: What Are the Different Types? BrainLine. https://www.brainline.org/slideshow/brain-imaging-what-are-different-types

[3] brainmap.org | GingerALE. (n.d.). Www.brainmap.org. https://www.brainmap.org/ale/

[4] Bunnell, D. (2024). Eating Disorder Statistics. National Eating Disorders Association. https://www.nationaleatingdisorders.org/statistics/

[5] citationchaser. (n.d.). Estech.shinyapps.io. https://estech.shinyapps.io/citationchaser/

[6] Cleveland Clinic. (2024). Eating disorders: Anorexia. Cleveland Clinic; Cleveland Clinic. https://my.clevelandclinic.org/health/diseases/9794-anorexia-nervosa

[7] Cowden, S. (2022, February 27). How Is Anorexia Nervosa Diagnosed? Verywell Mind. https://www.verywellmind.com/diagnostic-criteria-for-anorexia-nervosa-1138312

[8] Discovery Contributor. (2019, February 21). Expectations After Treatment: Statistics on Eating Disorders. Center for Discovery. https://centerfordiscovery.com/blog/statistics-on-eating-disorders/

[9] Eating Disorder Hope. (2019). Anorexia Nervosa – Highest Mortality Rate of Any Mental Disorder: Why? Eating Disorder Hope. https://www.eatingdisorderhope.com/information/anorexia/anorexia-death-rate

[10] Eating Disorder Statistics | ANAD - National Association of Anorexia Nervosa and Associated Disorders. (2023, November 29). ANAD - National Association of Anorexia Nervosa and Associated Disorders. https://anad.org/learning-library/eating-disorder-statistic/

[11] Eating Disorders in Midlife and Beyond- National Eating Disorders Association. (2025, November 10). National Eating Disorders Association. https://www.nationaleatingdisorders.org/eating-disorders-and-midlife-and-beyond

[12] Guy-Evans, O. (2021, June 9). Gyri and Sulci of the Brain - Simply Psychology. Www.simplypsychology.org. https://www.simplypsychology.org/gyri-and-sulci-of-the-brain.html

[13] Leigh, S. (2019, November 19). Many Patients with Anorexia Nervosa Get Better, But Complete Recovery Elusive to Most. Many Patients with Anorexia Nervosa Get Better, but Complete Recovery Elusive to Most | UC San Francisco. https://www.ucsf.edu/news/2019/11/416006/many-patients-anorexia-nervosa-get-better-complete-recovery-elusive-most

[14] PRISMA. (2020). Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA). PRISMA. https://www.prisma-statement.org/

[15] Research Imaging Institute — Mango. (n.d.). Mangoviewer.com. https://mangoviewer.com/

[16] Schroeder. (2025). Learn Meta-Analysis LLC. Learn Meta-Analysis LLC. https://learnmeta-analysis.com/

[17] Statistics on Anorexia in Men and Boys | Bulimia.com. (n.d.). Https://Bulimia.com/. https://bulimia.com/anorexia/statistics/men-boys/

[18] Su, T., Gong, J., Tang, G., Qiu, S., Chen, P., Chen, G., Wang, J., Huang, L., & Wang, Y. (2021). Structural and functional brain alterations in anorexia nervosa:A multimodal meta‐analysis of neuroimaging studies. Human Brain Mapping, 42(15), 5154–5169. https://doi.org/10.1002/hbm.25602

[19] Titova, O. E., Hjorth, O. C., Schiöth, H. B., & Brooks, S. J. (2013). Anorexia nervosa is linked to reduced brain structure in reward and somatosensory regions: a meta-analysis of VBM studies. BMC Psychiatry, 13(1). https://doi.org/10.1186/1471-244x-13-110

[20] Welcome to Andy’s Brain Book! — Andy’s Brain Book 1.0 documentation. (n.d.). Andysbrainbook.readthedocs.io. https://andysbrainbook.readthedocs.io/en/latest/index.html

Included Articles in Analysis

[1] Bär, K.-J., de la Cruz, F., Berger, S., Schultz, C. C., & Wagner, G. (2015). Structural and functional differences in the cingulate cortex relate to disease severity in anorexia nervosa. Journal of Psychiatry & Neuroscience, 40(4), 269–279. https://doi.org/10.1503/jpn.140193

[2] Björnsdotter, M., Davidovic, M., Karjalainen, L., Starck, G., Olausson, H., & Wentz, E. (2018). Grey matter correlates of autistic traits in women with anorexia nervosa. Journal of Psychiatry & Neuroscience, 43(2), 79–86. https://doi.org/10.1503/jpn.170072

[3] Boehm, I., Geisler, D., King, J. A., Ritschel, F., Seidel, M., Deza Araujo, Y., Petermann, J., Lohmeier, H., Weiss, J., Walter, M., Roessner, V., & Ehrlich, S. (2014). Increased resting state functional connectivity in the fronto-parietal and default mode network in anorexia nervosa. Frontiers in Behavioral Neuroscience, 8. https://doi.org/10.3389/fnbeh.2014.00346

[4] Boehm, I., Geisler, D., Tam, F., King, J. A., Ritschel, F., Seidel, M., Bernardoni, F., Murr, J., Goschke, T., Calhoun, V. D., Roessner, V., & Ehrlich, S. (2016). Partially restored resting-state functional connectivity in women recovered from anorexia nervosa. Journal of Psychiatry & Neuroscience, 41(6), 377–385. https://doi.org/10.1503/jpn.150259

[5] Bomba, M., Riva, A., Morzenti, S., Grimaldi, M., Neri, F., & Nacinovich, R. (2015). Global and regional brain volumes normalization in weight-recovered adolescents with anorexia nervosa: preliminary findings of a longitudinal voxel-based morphometry study. Neuropsychiatric Disease and Treatment, 637. https://doi.org/10.2147/ndt.s73239

[6] Brodrick, B. B., Adler-Neal, A. L., Palka, J. M., Mishra, V., Aslan, S., & McAdams, C. J. (2021). Structural brain differences in recovering and weight-recovered adult outpatient women with anorexia nervosa. Journal of Eating Disorders, 9(1). https://doi.org/10.1186/s40337-021-00466-w

[7] Brooks, S. J., Barker, G. J., O’Daly, O. G., Brammer, M., Williams, S. C., Benedict, C., Schiöth, H. B., Treasure, J., & Campbell, I. C. (2011). Restraint of appetite and reduced regional brain volumes in anorexia nervosa: a voxel-based morphometric study. BMC Psychiatry, 11(1). https://doi.org/10.1186/1471-244x-11-179

[8] Collantoni, E., Meneguzzo, P., Tenconi, E., Manara, R., & Favaro, A. (2019). Small-world properties of brain morphological characteristics in Anorexia Nervosa. PLOS ONE, 14(5), e0216154. https://doi.org/10.1371/journal.pone.0216154

[9] Cowdrey, F. A., Filippini, N., Park, R. J., Smith, S. M., & McCabe, C. (2012). Increased resting state functional connectivity in the default mode network in recovered anorexia nervosa. Human Brain Mapping, 35(2), 483–491. https://doi.org/10.1002/hbm.22202

[10] de la Cruz, F., Schumann, A., Suttkus, S., Helbing, N., Zopf, R., & Bär, K.-J. (2021). Cortical thinning and associated connectivity changes in patients with anorexia nervosa. Translational Psychiatry, 11(1). https://doi.org/10.1038/s41398-021-01237-6

[11] Doose, A., Tam, F. I., Hellerhoff, I., King, J. A., Boehm, I., Gottloeber, K., Wahl, H., Werner, A., Raschke, F., Bartnik-Olson, B., Lin, A. P., Akgün, K., Roessner, V., Linn, J., & Ehrlich, S. (2023). Triangulating brain alterations in anorexia nervosa: a multimodal investigation of magnetic resonance spectroscopy, morphometry and blood-based biomarkers. Translational Psychiatry, 13(1), 1–8. https://doi.org/10.1038/s41398-023-02580-6

[12] Ehrlich, S., Lord, A. R., Geisler, D., Borchardt, V., Boehm, I., Seidel, M., Ritschel, F., Schulze, A., King, J. A., Weidner, K., Roessner, V., & Walter, M. (2015). Reduced functional connectivity in the thalamo-insular subnetwork in patients with acute anorexia nervosa. Human Brain Mapping, 36(5), 1772–1781. https://doi.org/10.1002/hbm.22736

[13] Favaro, A., Tenconi, E., Degortes, D., Manara, R., & Santonastaso, P. (2015). Gyrification brain abnormalities as predictors of outcome in anorexia nervosa. Human Brain Mapping, 36(12), 5113–5122. https://doi.org/10.1002/hbm.22998

[14] Fujisawa, T. X., Yatsuga, C., Mabe, H., Yamada, E., Masuda, M., & Tomoda, A. (2015). Anorexia Nervosa during Adolescence Is Associated with Decreased Gray Matter Volume in the Inferior Frontal Gyrus. PLOS ONE, 10(6), e0128548. https://doi.org/10.1371/journal.pone.0128548

[15] Gaudio, S., Olivo, G., Beomonte Zobel, B., & Schiöth, H. B. (2018). Altered cerebellar–insular–parietal–cingular subnetwork in adolescents in the earliest stages of anorexia nervosa: a network–based statistic analysis. Translational Psychiatry, 8(1). https://doi.org/10.1038/s41398-018-0173-z

[16] Gaudio, S., Piervincenzi, C., Beomonte Zobel, B., Romana Montecchi, F., Riva, G., Carducci, F., & Cosimo Quattrocchi, C. (2015). Altered resting state functional connectivity of anterior cingulate cortex in drug naïve adolescents at the earliest stages of anorexia nervosa. Scientific Reports, 5(1). https://doi.org/10.1038/srep10818

[17] Hu, S., Feng, H., Xu, T., Zhang, H., Zhao, Z., Lai, J., Xu, D., & Xu, Y. (2017). Altered microstructure of brain white matter in females with anorexia nervosa: a diffusion tensor imaging study. Neuropsychiatric Disease and Treatment, Volume 13, 2829–2836. https://doi.org/10.2147/ndt.s144972

[18] Kullmann, S., Giel, K. E., Teufel, M., Thiel, A., Zipfel, S., & Preissl, H. (2014). Aberrant network integrity of the inferior frontal cortex in women with anorexia nervosa. NeuroImage: Clinical, 4, 615–622. https://doi.org/10.1016/j.nicl.2014.04.002

[19] Lai, J., Xu, T., Zhang, H., Xi, C., Zhu, H., Du, Y., Jiang, J., Wu, L., Zhang, P., Xu, Y., Hu, S., & Xu, D. (2020, February 28). Fractional amplitude of low frequency fluctuation in drug-naïve first-episode patients with anorexia nervosa (M. Tusconi, Ed.). Medicine; Lippincott. https://journals.lww.com/md-journal/Fulltext/2020/02280/Fractional_amplitude_of_low_frequency_fluctuation.47.aspx

[20] Leppanen, J., Sedgewick, F., Cardi, V., Treasure, J., & Tchanturia, K. (2019). Cortical morphometry in anorexia nervosa: An out‐of‐sample replication study. European Eating Disorders Review. https://doi.org/10.1002/erv.2686

[21] Mainz, V., Schulte-Rüther, M., Fink, G. R., Herpertz-Dahlmann, B., & Konrad, K. (2012). Structural Brain Abnormalities in Adolescent Anorexia Nervosa Before and After Weight Recovery and Associated Hormonal Changes. Psychosomatic Medicine, 74(6), 574–582. https://doi.org/10.1097/psy.0b013e31824ef10e

[22] McFadden, K., Tregellas, J., Shott, M., & Frank, G. (2014). Reduced salience and default mode network activity in women with anorexia nervosa. Journal of Psychiatry & Neuroscience, 39(3), 178–188. https://doi.org/10.1503/jpn.130046

[23] Mühlau, M., Gaser, C., Ilg, R., Conrad, B., Leibl, C., Cebulla, M. H., Backmund, H., Gerlinghoff, M., Lommer, P., Schnebel, A., Wohlschläger, A. M., Zimmer, C., & Nunnemann, S. (2007). Gray Matter Decrease of the Anterior Cingulate Cortex in Anorexia Nervosa. American Journal of Psychiatry, 164(12), 1850–1857. https://doi.org/10.1176/appi.ajp.2007.06111861

[24] Myrvang, A. D., Vangberg, T. R., Stedal, K., Rø, Ø., Endestad, T., Rosenvinge, J. H., & Aslaksen, P. M. (2020). Cerebral cortical thickness and surface area in adolescent anorexia nervosa: Separate and joint analyses with a permutation‐based nonparametric method. International Journal of Eating Disorders, 54(4), 561–568. https://doi.org/10.1002/eat.23448

[25] Olivo, G., Swenne, I., Zhukovsky, C., Tuunainen, A.-K., Saaid, A., Salonen-Ros, H., Larsson, E.-M., Brooks, S. J., & Schiöth, H. B. (2019). Preserved white matter microstructure in adolescent patients with atypical anorexia nervosa. International Journal of Eating Disorders, 52(2), 166–174. https://doi.org/10.1002/eat.23012

[26] Olivo, G., Swenne, I., Zhukovsky, C., Tuunainen, A.-K., Salonen-Ros, H., Larsson, E.-M., Gaudio, S., Brooks, S. J., & Schiöth, H. B. (2018). Reduced resting-state connectivity in areas involved in processing of face-related social cues in female adolescents with atypical anorexia nervosa. Translational Psychiatry, 8(1). https://doi.org/10.1038/s41398-018-0333-1

[27] Pfuhl, G., King, J. A., Geisler, D., Roschinski, B., Ritschel, F., Seidel, M., Bernardoni, F., Müller, D. K., White, T., Roessner, V., & Ehrlich, S. (2016). Preserved white matter microstructure in young patients with anorexia nervosa? Human Brain Mapping, 37(11), 4069–4083. https://doi.org/10.1002/hbm.23296

[28] Travis, K. E., Golden, N. H., Feldman, H. M., Solomon, M., Nguyen, J., Mezer, A., Yeatman, J. D., & Dougherty, R. F. (2015). Abnormal white matter properties in adolescent girls with anorexia nervosa. NeuroImage: Clinical, 9, 648–659. https://doi.org/10.1016/j.nicl.2015.10.008

[29] Uniacke, B., Wang, Y., Biezonski, D., Sussman, T., Lee, S., Posner, J., & Steinglass, J. (2018). Resting-state connectivity within and across neural circuits in anorexia nervosa. Brain and Behavior, 9(1), e01205. https://doi.org/10.1002/brb3.1205

[30] Via, E., Zalesky, A., Sánchez, I., Forcano, L., Harrison, B. J., Pujol, J., Fernández-Aranda, F., Menchón, J. M., Soriano-Mas, C., Cardoner, N., & Fornito, A. (2014). Disruption of brain white matter microstructure in women with anorexia nervosa. Journal of Psychiatry and Neuroscience, 39(6), 367–375. https://doi.org/10.1503/jpn.130135

[31] Vogel, K., Timmers, I., Kumar, V., Nickl-Jockschat, T., Bastiani, M., Roebroek, A., Herpertz-Dahlmann, B., Konrad, K., Goebel, R., & Seitz, J. (2016). White matter microstructural changes in adolescent anorexia nervosa including an exploratory longitudinal study. NeuroImage: Clinical, 11, 614–621. https://doi.org/10.1016/j.nicl.2016.04.002

[32] Yonezawa, H., Otagaki, Y., Miyake, Y., Okamoto, Y., & Yamawaki, S. (2008). No differences are seen in the regional cerebral blood flow in the restricting type of anorexia nervosa compared with the binge eating/purging type. Psychiatry and Clinical Neurosciences, 62(1), 26–33. https://doi.org/10.1111/j.1440-1819.2007.01769.x

[33] Zhang, S., Wang, W., Su, X., Kemp, G. J., Yang, X., Su, J., Tan, Q., Zhao, Y., Sun, H., Yue, Q., & Gong, Q. (2018). Psychoradiological investigations of gray matter alterations in patients with anorexia nervosa. Translational Psychiatry, 8. https://doi.org/10.1038/s41398-018-0323-3