Breathlessness catastrophising after COVID-19 involves both interoceptive and visceromotor connectivity
Qiu, L., Raman, B., Cassar, M., Fraser, E., Atkinson, L. Z., Clare, S., Clarke, W. T., Neubauer, S., et al.
neuroscience
Quick verdict
7T fMRI in 53 post-COVID patients ties breathlessness catastrophising to dPAG-insula and BLA-dACC connectivity; rigorous but small, with site-confounded moderation.
Summary
This preprint asks why post-COVID breathlessness correlates poorly with lung physiology. Fifty-three patients (36% hospitalised, 14 previously mechanically ventilated; median 257 days post-infection) underwent 10-minute 7-tesla resting-state fMRI. Connectivity among 18 pre-defined interoceptive/visceromotor ROIs (153 edges) was regressed on the Breathlessness Catastrophising Scale (BCS) and Dyspnoea-12 (D-12) using robust (Huber) regression with demographic/imaging covariates and FDR correction per outcome. Two BCS edges survived: negative dPAG-PoI1 connectivity (beta=-4.62) and positive BLA-dACC connectivity (beta=4.14). Exploratory moderation showed the dPAG-PoI1 association stronger in ventilated patients (permutation p=.020) and the BLA-dACC association attenuated at higher GAD-7 (p=.007); formal cross-edge bootstrap contrasts included zero, so no statistical dissociation is claimed. No D-12 edge survived. Effects were stable across six sensitivity specifications adding site, cohort, WHO severity, time since infection, and respiratory history. Main limitations: N=53 with ventilation structurally confounded with scanning site (all ventilated patients at Oxford), right-skewed GAD-7, cross-sectional design, selected-edge-only diagnostics, limited pulmonary phenotyping, and no stated code/data availability.
Review complete
Assessment at a glance
5
risk flags identified
● high 0
● medium 4
● low 1
5 strengths · 5 weaknesses · 5 author questions
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Breathlessness catastrophising after COVID-19 is associated with reduced resting-state connectivity between the dorsal periaqueductal grey and posterior insula (dPAG-PoI1, beta=-4.62, FDR-corrected p=.0062).
- Breathlessness catastrophising is associated with increased connectivity between the basolateral amygdala and dorsal anterior cingulate (BLA-dACC, beta=4.14, FDR-corrected).
- The dPAG-PoI1 association is stronger in previously mechanically ventilated patients while the BLA-dACC association is attenuated at higher generalised anxiety, suggesting partially separable interoceptive and visceromotor contributions.
- No functional connectivity association with the broader Dyspnoea-12 measure survived correction, and post-COVID breathlessness is proposed to reflect a closed-loop perception-action disturbance rather than anxiety-driven misperception alone.
Strengths
Hypothesis-constrained analytic space
Inference is limited to 18 a priori ROIs (153 edges) from the Kleckner et al. (2017) interoceptive-allostatic framework rather than a whole-brain search, sharply reducing analytic flexibility; ROI provenance (Glasser PoI1/PoI2, Kleckner 4-mm spheres, FreeSurfer hypothalamic subunits) is documented in Methods.
Statistical pipeline goes beyond the typical correlation matrix
Robust (Huber) regression with FDR control per outcome, leave-one-out and BCa bootstrap (5,000 resamples) diagnostics, DFBETAS screening, permutation tests for interaction terms, and a pre-specified decision rule for the cross-edge contrast - the authors openly reject their own double-dissociation interpretation when the CIs include zero (Results, 'Formal test of differential moderation').
Acquisition and denoising suited to the hardest targets
Unusually transparent confound reporting
Honest null result for the second outcome
Weaknesses
Structural confound sits exactly where the narrative needs it
Winner's-curse exposure on selected edges
The anxiety moderation rests on a thin tail of the moderator
Outcome measures are self-report with limited post-COVID validation
No stated data or code availability
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 5 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What did this study find about breathlessness after COVID-19?
In 53 post-COVID patients scanned with 7-tesla resting-state fMRI, greater breathlessness catastrophising was associated with weaker connectivity between the dorsal periaqueductal grey and posterior insula and stronger connectivity between the basolateral amygdala and dorsal anterior cingulate. The authors interpret this as combined interoceptive (signal-sampling) and visceromotor (threat-regulation) contributions to the symptom, rather than anxiety-driven misperception alone.
Is long-COVID breathlessness just anxiety?
This paper argues no. The brainstem-insula connection linked to catastrophising showed little moderation by generalised anxiety (GAD-7), and the amygdala-cingulate association was actually weakest in the most anxious patients. Anxiety appears to modulate rather than explain the brain-behaviour associations, though the study is cross-sectional and cannot establish causation.
How large was the study and what are its main caveats?
53 patients (14 previously mechanically ventilated), scanned a median of 257 days after infection at two UK 7T sites. Main caveats: small sample, ventilation status fully confounded with scanning site and hospitalisation, cross-sectional design, self-report outcomes, exploratory moderation analyses on pre-selected edges, and no surviving association with the broader Dyspnoea-12 measure.
Did the study prove the two brain connections reflect different mechanisms?
No. The moderation patterns pointed in a double-dissociation direction, but the formal bootstrap contrasts between the two edges included zero, and the authors explicitly state they do not claim a statistical dissociation - they describe the two contributions as 'potentially separable' and frame the loop model as an interpretive framework, not a demonstrated causal architecture.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Appendices 1-8 (ROI coordinates, QC metrics, full covariate table, cohort detail) are referenced but not included in the extracted text, so ROI definitions and QC results could not be verified.
- Figures are available only through captions; matrix values, scatter distributions, and moderation plots could not be cross-checked against the numbers in the text.
- No data or code availability statement appears in the extracted text; one may exist in a section not captured by the extraction.
- Several listed references (e.g., Stoeckel et al. 2018; Mitchell & Baker 2022; Sullivan et al. 1995) do not appear cited in the main text and are presumably cited only in the non-included appendices.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.