Acute blood biomarker profiles predict cognitive deficits 6 and 12 months after COVID-19 hospitalization

Post-COVID cognitive deficits, including ‘brain fog’, are clinically complex, with both objective and subjective components. They are common and debilitating, and can affect the ability to work, yet their biological underpinnings remain unknown. In this prospective cohort study of 1,837 adults hospitalized with COVID-19, we identified two distinct biomarker profiles measured during the acute admission, which predict cognitive outcomes 6 and 12 months after COVID-19. A first profile links elevated fibrinogen relative to C-reactive protein with both objective and subjective cognitive deficits. A second profile links elevated D-dimer relative to C-reactive protein with subjective cognitive deficits and occupational impact. This second profile was mediated by fatigue and shortness of breath. Neither profile was significantly mediated by depression or anxiety. Results were robust across secondary analyses. They were replicated, and their specificity to COVID-19 tested, in a large-scale electronic health records dataset. These findings provide insights into the heterogeneous biology of post-COVID cognitive deficits.

One possibility would be that dimensions of covariation imply associations at the univariate level (i.e. that high fibrinogen and high D-dimer themselves predict cognitive outcomes).However, we found this not to be the case.Adjusting for all covariates as in the primary analysis, fibrinogen was significantly associated with neither C-PSQ nor MoCA at 6-months (p>0.05 for both), and neither was D-dimer (p>0.05 for both).One reason might be that such univariate associations fail to encode the role of CRP in the biomarker profile.
A simplified representation of the dimensions of covariation which captures the role of CRP is to simplify the biomarker profile to be the difference between fibrinogen and CRP for the first dimension of covariation and the difference between D-dimer and CRP for the second dimension of covariation (more specifically the difference between covariate-adjusted and, for D-dimer and CRP, log-transformed biomarker values).This amounts to setting the weights of the weighted combinations representing biomarker profiles to 1 for fibrinogen (respectively D-dimer for the second dimension) and -1 for CRP, while setting all other weights to zero.These simplified biomarker profiles were found to be significantly associated with C-PSQ at 6 months (r=0.09and p<0.0001 for both dimensions).This implies that simplified biomarker profiles in which only fibrinogen and CRP (respectively D-dimer and CRP for the second dimension) are kept and in which their relative contributions are given equal weights are enough to capture the association between biomarkers and subjective cognitive deficit.Interestingly, the simplified profile for the first dimension of covariation was not significantly associated with the MoCA at 6-months, and the full dimension of covariation was needed to

Supplementary Note 4
The effect of confounding in the replication study with EHR data was reduced with 1:1 matching on the following factors (provided with ICD-10/CDC codes in brackets): 1) Age at the time of index event.
2) Sex coded as female, male, or other.2B of the main manuscript).The correlations and p-values are also provided for the association between cognitive outcomes and the dimension of covariation (i.e. the location of individuals along the dimension of covariation which is calculated as the mean between the cognitive and biomarker profile).Two-sided p-values (not adjusted for multiple comparisons) for each correlation were calculated using a permutation test.Note that the minimum p-value achievable with a permutation test with 10,000 permutations is 0.0001.Some associations might have smaller p-values and would be reported as p=0.0001  3B of the main manuscript).The correlations and p-values are also provided for the association between cognitive outcomes and the dimension of covariation (i.e. the location of individuals along the dimension of covariation which is calculated as the mean between the cognitive and biomarker profile).Two-sided p-values (not adjusted for multiple comparisons) for each correlation were calculated using a permutation test.Note that the minimum p-value achievable with a permutation test with 10,000 permutations is 0.0001.Some associations might have smaller p-values and would be reported as p=0.0001

Rehabilitation,
Sarcopenia and Fatigue S J Singh (Co-Lead), W D-C Man (Co-Lead), J M Lord (Co-Lead), N J Greening (Co-Lead), T Chalder (Co-Lead), J T Scott (Co-Lead), N Armstrong, E Baldry, M Baldwin, N Basu, M Beadsworth, L Bishop, C E Bolton, A Briggs, M Buch, G Carson, J Cavanagh, H Chinoy, C Dawson, E Daynes, S Defres, R A Evans, L Gardiner, P Greenhaff, S Greenwood, M Harvie, L HOuchen-Wolloff, M Husain, S MacDonald, A McArdle, H J C McAuley, A McMahon, M McNarry, G Mills, C Nolan, K O'Donnell, D Parekh, Pimm, J Sargent, L Sigfrid, M Steiner, D Stensel, A L Tan, I Vogiatzis, J Whitney, D Wilkinson, D Wilson, M Witham, D G Wootton, T Yates Renal D Thomas (Lead), N Brunskill (Co-Lead), S Francis (Co-Lead), S Greenwood (Co-Lead), C Laing (Co-Lead), K Bramham, P Chowdhury, A Frankel, L Lightstone, S McAdoo, K McCafferty, M Ostermann, N Selby, C Sharpe, M Willicombe Patient Public Engagement Group L Houchen-Wolloff (Lead), J Bunker, R Gill, C Hastie, R Nathu, N Rogers, N Smith Local Clinical Centre PHOSP-COVID trial staff (listed in alphabetical order) Airedale NHS Foundation Trust A Shaw (PI), L Armstrong, B Hairsine, H Henson, C Kurasz, L Shenton Aneurin Bevan University Health Board S Fairbairn (PI), A Dell, N Hawkings, J Haworth, M Hoare, A Lucey, V Lewis, G Mallison, H Nassa, C Pennington, A Price, C Price, A Storrie, G Willis, S Young Barts Health NHS Trust & Queen Mary University of London P Pfeffer (PI), K Chong-James, C David, W Y James, C Manisty, A Martineau, O Zongo Barnsley Hospital NHS Foundation Trust A Sanderson (PI) Belfast Health and Social Care Trust & Queen's University Belfast L G Heaney (PI), C Armour, V Brown, T Craig, S Drain, B King, N Magee, D McAulay, E Major, L McGarvey, J McGinness, R Stone Betsi Cadwaladr University Health Board A Haggar (PI), A Bolger, F Davies, J Lewis, A Lloyd, R Manley, E McIvor, D Menzies, K Roberts, W Saxon, D Southern, C Subbe, V Whitehead Borders General Hospital, NHS Borders H El-Taweel (PI), J Dawson, L Robinson Bradford Teaching Hospitals NHS Foundation Trust D Saralaya (PI), L Brear, K Regan, K Storton Cambridge University Hospitals NHS Foundation Trust, NIHR Cambridge Clinical Research Facility & University of Cambridge J Fuld (PI), A Bermperi, I Cruz, K Dempsey, A Elmer, H Jones, S Jose, S Marciniak, M Parkes, C Ribeiro, J Taylor, M Toshner, L Watson, J Weir McCall, J Worsley Cardiff and Vale University Health Board R Sabit (PI), L Broad, A Buttress, T Evans, M Haynes, L Jones, L Knibbs, A McQueen, C Oliver, K Paradowski, J Williams Chesterfield Royal Hospital NHS Trust E Harris (PI), C Sampson Cwm Taf Morgannwg University Health Board C Lynch (PI), E Davies, C Evenden , A Hancock, K Hancock, M Rees , L Roche, N Stroud, T Thomas-Woods East Cheshire NHS Trust M Babores (PI), J Bradley-Potts, M Holland, N Keenan, S Shashaa, H Wassall East Kent Hospitals University NHS Foundation Trust E Beranova (PI), H Weston (PI), T Cosier, L Austin, J Deery, T Hazelton, C Price, H Ramos, R Solly, S Turney Gateshead NHS Trust L Pearce (PI), W McCormick, S Pugmire, W Stoker, A Wilson Guy's and St Thomas' NHS Foundation Trust N Hart (PI), LA Aguilar Jimenez, G Arbane, S Betts, K Bisnauthsing, A Dewar, P Chowdhury, A Chiribiri, A Dewar, G Kaltsakas, H Kerslake, MM Magtoto, P Marino, LM Martinez, C O'Brien, M Ostermann, J Rossdale, TS Solano, E Wynn Hampshire Hospitals NHS Foundation Trust N Williams (PI), W Storrar (PI), M Alvarez Corral, A Arias, E Bevan, D Griffin, J Martin, J Owen, S Payne, A Prabhu, A Reed, C Wrey Brown Harrogate and District NHD Foundation Trust C Lawson (PI), T Burdett, J Featherstone, A Layton, C Mills, L Stephenson, Hull University Teaching Hospitals NHS Trust & University of Hull N Easom (PI), P Atkin, K Brindle, M G Crooks, K Drury, R Flockton, L Holdsworth, A Richards, D L Sykes, S Thackray-Nocera, C Wright Hywel Dda University Health Board K E Lewis (PI), A Mohamed (PI), G Ross (PI), S Coetzee, K Davies, R Hughes, R Loosley, L O'Brien, Z Omar, H McGuinness, E Perkins, J Phipps, A Taylor, H Tench, R Wolf-Roberts Imperial College Healthcare NHS Trust & Imperial College London L S Howard (PI), O Kon (PI), D C Thomas (PI), S Anifowose, L Burden, E Calvelo, B Card, C Carr, E R Chilvers, D Copeland, P Cullinan, P Daly, L Evison, T Fayzan, H Gordon, S Haq, R G Jenkins, C King, K March, M Mariveles, L McLeavey, N Mohamed, S Moriera, U Munawar, J Nunag, U Nwanguma, L Orriss-Dib, D P O'Regan, A Ross, M Roy, E Russell, K Samuel, J Schronce, N Simpson, L Tarusan, C Wood, N Yasmin Kettering General Hospital NHS Trust R Reddy (PI), A-M, Guerdette, M Hewitt, K Warwick, S White King's College Hospital NHS Foundation Trust & Kings College London A M Shah (PI), C J Jolley (PI), O Adeyemi, R Adrego, H Assefa-Kebede, J Breeze, M Brown, S Byrne, T Chalder, A Chiribiri, P Dulawan, N Hart, A Hayday, A Hoare, A Knighton, M Malim, C O'Brien, S Patale, I Peralta, N Powell, A Ramos, K Shevket, F Speranza, A Te Leeds Teaching Hospitals & University of Leeds P Beirne (PI), A Ashworth, J Clarke, C Coupland, M Dalton, E Wade, C Favager, J Greenwood, J Glossop, L Hall, T Hardy, A Humphries, J Murira, D Peckham, S Plein, J Rangeley, G Saalmink, A L Tan, B Whittam, N Window, J Woods, Lewisham & Greenwich NHS Trust G Coakley (PI) Liverpool University Hospitals NHS Foundation Trust & University of Liverpool D G Wootton (PI), L Turtle (PI), L Allerton, AM All, M Beadsworth, A Berridge, J Brown, S Cooper, A Cross, D J Cuthbertson, S Defres, S L Dobson, J Earley, N French, W Greenhalf, H E Hardwick, K Hainey, J Hawkes, V Highett, S Kaprowska, G J Kemp, AL Key, S Koprowska, L Lavelle-Langham, N Lewis-Burke, G Madzamba, F Malein, S Marsh, C Mears, L Melling, M J Noonan, L Poll, J Pratt, E Richardson, A Rowe, M G Semple, V Shaw, K A Tripp, B Vinson, L O Wajero, S A Williams-Howard, J Wyles London North West University Healthcare NHS Trust S N Diwanji (PI), P Papineni (PI), S Gurram, S Quaid, G F Tiongson, E Watson Manchester University NHS Foundation Trust & University of Manchester B Al-Sheklly (PI), A Horsley (PI), C Avram, J Blaikely, M Buch, N Choudhury, D Faluyi, T Felton, T Gorsuch, N A Hanley, T Hussell, Z Kausar, C A Miller, N Odell, R Osbourne, K Piper Hanley, K Radhakrishnan, S Stockdale Newcastle upon Tyne Hospitals NHS Foundation Trust & University of Newcastle A De Soyza (PI), C Echevarria (PI), A Ayoub, J Brown, G Burns, G Davies, H Fisher, C Francis, A Greenhalgh, P Hogarth, J Hughes, K Jiwa, G Jones, G MacGowan, D Price, A Sayer, J Simpson, H Tedd, S Thomas, S West, M Witham, S Wright, A Young NHS Dumfries and Galloway M J McMahon (PI), P Neill NHS Greater Glasgow and Clyde Health Board & University of Glasgow D Anderson (PI), H Bayes (PI), C Berry (PI), D Grieve (PI), I B McInnes (PI), N Basu, A Brown, A Dougherty, K Fallon, L Gilmour, K Mangion, A Morrow, K Scott, R Sykes, R Touyz NHS Highland E K Sage (PI), F Barrett, A Donaldson NHS Lanarkshire M Patel (PI), D Bell, A Brown, M Brown, R Hamil, K Leitch, L Macliver, J Quigley, A Smith, B Welsh NHS Lothian & University of Edinburgh St George's University Hospitals NHS Foundation Trust R Aul (PI), M Ali, A Dunleavy (PI), D Forton, N Msimanga, M Mencias, T Samakomva, S Siddique, J Teixeira, V Tavoukjian Sherwood Forest Hospitals NHS Foundation Trust J Hutchinson (PI), L Allsop, K Bennett, P Buckley, M Flynn, M Gill, C Goodwin, M Greatorex, H Gregory, C Heeley, L Holloway, M Holmes, J Kirk, W Lovegrove, TA Sewell, S Shelton, D Sissons, K Slack, S Smith, D Sowter, S Turner, V Whitworth, I Wynter Shropshire Community Health NHS Trust L Warburton (PI), S Painter, J Tomlinson Somerset NHS Foundation Trust C Vickers (PI), T Wainwright, D Redwood, J Tilley, S Palmer Swansea Bay University Health Board G A Davies (PI), L Connor, A Cook, T Rees, F Thaivalappil, C Thomas Tameside and Glossop Integrated Care NHS Foundation A Butt (PI), M Coulding, H Jones, S Kilroy, J McCormick, J McIntosh, H Savill, V Turner, J Vere The Great Western Hospital Foundation Trust E Fraile (PI), J Ugoji The Hillingdon Hospitals NHS Foundation Trust S S Kon (PI), H Lota, G Landers, M Nasseri, S Portukhay The Rotherham NHS Foundation Trust A Hormis (PI), A Daniels, J Ingham, L Zeidan United Lincolnshire Hospitals NHS Trust M Chablani (PI), L Osborne University College London Hospital & University College London M Marks (PI), J S Brown (PI), N Ahwireng, B Bang, D Basire, R C Chambers, A Checkley, R Evans, M Heightman, T Hillman, J Hurst, J Jacob, S Janes, R Jastrub, M Lipman, S Logan, D Lomas, M Merida Morillas, A Pakzad, H Plant, J C Porter, K Roy, E Wall, B Williams, M Xu University Hospital Birmingham NHS Foundation Trust & University of Birmingham D Parekh (PI), N Ahmad Haider, C Atkin, R Baggott, M Bates, A Botkai, A Casey, B Cooper, J Dasgin, K Draxlbauer, N Gautam, J Hazeldine, T Hiwot, S Holden, K Isaacs, T Jackson, S Johnson, V Kamwa, D Lewis, J M Lord, S Madathil, C McGhee, K Mcgee, A Neal, A Newton Cox, J Nyaboko, D Parekh, Z Peterkin, H Qureshi, B Rangelov, L Ratcliffe, E Sapey, J Short, T Soulsby, R Steeds, J Stockley, Z Suleiman, T Thompson, M Ventura, S Walder, C Welch, D Wilson, S Yasmin, K P Yip University Hospitals of Derby and Burton P Beckett (PI) C Dickens, U Nanda University Hospitals of Leicester NHS Trust & University of Leicester C E Brightling (CI), R A Evans (PI), M Aljaroof, N Armstrong, H Arnold, H Aung, M Bakali, M Bakau, M Baldwin, M Bingham, M Bourne, C Bourne, N Brunskill, P Cairns, L Carr, A Charalambou, C Christie, M J Davies, S Diver, S Edwards, C Edwardson, O Elneima, H Evans, J Finch, S Glover, N Goodman, B Gootpu, N J Greening, K Hadley, P Haldar, B Hargadon, V C Harris, L Houchen-Wolloff, W Ibrahim, L Ingram, K Khunti, A Lea, D Lee, G P McCann, H J C McAuley, P McCourt, T Mcnally, G Mills, A Moss, W Monteiro, M Pareek, S Parker, A Rowland, A Prickett, I N Qureshi, R Russell, N Samani, M Sereno, M Sharma, A Shikotra, S Siddiqui, A Singapuri, S J Singh, J Skeemer, M Soares, E Stringer, T Thornton, M Tobin, E Turner, L V Wain, T J C Ward, F Woodhead, J Wormleighton, T Yates, A Yousuf University Hospital Southampton NHS Foundation Trust & University of Southampton M G Jones (PI), C Childs, R Djukanovic, S Fletcher, M Harvey, E Marouzet, B Marshall, R Samuel, T Sass, T Wallis, H Wheeler Whittington Health NHS R Dharmagunawardena (PI), E Bright, P Crisp, M Stern Wirral University Teaching Hospital A Wight (PI), L Bailey, A Reddington Wrightington Wigan and Leigh NHS trust A Ashish (PI), J Cooper, E Robinson Yeovil District Hospital NHS Foundation Trust A Broadley (PI) York & Scarborough NHS Foundation Trust K Howard (PI), L Barman, C Brookes, K Elliott.L Griffiths, Z Guy, D Ionita, H Redfearn, Asthma UK and British Lung Foundation Partnership -K Poinasamy, S Walker Royal Surrey NHS Foundation Trust M Halling-Brown South London and Maudsley NHS Foundation Trust & Kings College London G Breen, M Hotopf Swansea University & Swansea Welsh Network K Lewis, N Williams raised fibrinogen relative to CRP for the first dimension), such a simplified view might fail to capture the complexity of the biomarker profile necessary to predict cognitive outcomes.In this supplementary note, we explore the simplest representation of the dimension of covariation that can still reliably reproduce the main results from the manuscript.

o
Difficulty with concentration (Yes: +1 point, No: 0 point) o Difficulty with communication (Yes: +1 point, No: 0 point) o Slowing down in your thinking (Yes: +1 point, No: 0 point) § Do you have difficulty remembering or concentrating?: o No: 0 points o Yes, some difficulty: + 1/3 point o Yes, a lot of difficulty: + 2/3 points o Yes, cannot do at all: +1 point § Do you have difficulty communicating, for example understanding or being understood?o No: 0 points o Yes, some difficulty: + 1/3 point o Yes, a lot of difficulty: + 2/3 points o Yes, cannot do at all: +1 point

Table 2 :
Weights of the weighted combination representing the biomarker profile of the first dimension of covariation, relative weights (representing the correlation of each biomarker with the biomarker profile), and two-sided p-values (not adjusted for multiple comparisons) for each relative weight calculated using a permutation test.Note that the minimum p-value achievable with a permutation test with 10,000 permutations is 0.0001.Some associations might have smaller p-values and would be reported as p=0.0001.
Supplementary Table 3:Correlations and p-values (derived from permutation tests, two-sided, and not adjusted for multiple comparisons) between cognitive outcomes (including cognitive scores and signs of occupational impact at 6 and 12 months) and the cognitive profile of the first dimension (i.e.what is displayed in Fig.

Table 4 :
Weights of the weighted combination representing the biomarker profile of the second dimension of covariation, relative weights (representing the correlation of each biomarker with the biomarker profile), and two-sided p-values for each relative weight calculated using a permutation test (twosided and non-adjusted for multiple comparisons).Note that the minimum p-value achievable with a permutation test with 10,000 permutations is 0.0001.Some associations might have smaller p-values and would be reported as p=0.0001 Supplementary Table 5:Correlations and p-values (derived from permutation tests, two-sided, and not adjusted for multiple comparisons) between cognitive outcomes (including cognitive scores and signs of occupational impact at 6 and 12 months) and the cognitive profile of the second dimension of covariation (i.e.what is displayed in Fig.

Table 6 :
Results of the mediation analysis wherein the biomarker profile was the independent variable, the cognitive profile was the dependent variable, and mediators were 14 different clinical scales capturing different aspect of participants' health.Specifically, the table presents fractions explained by each mediator for the link between biomarker and cognitive profile alongside the 1 st and 2 nd mode of covariation, as well as the two-sided p-value testing the null hypothesis of no mediation.P-values were two-sided, not adjusted for multiple comparisons, and derived from non-parametric bootstrap with 1000 bootstrap repetitions.

Table 7 :
Baseline characteristics from the EHR dataset in the cohorts hospitalised with COVID-19, and with high and low fibrinogen (and normal CRP), before and after propensity score matching.

Table 8 :
Baseline characteristics from the EHR dataset in the cohorts hospitalised with COVID-19, and with high and low D-dimer, before and after propensity score matching.

Table 9 :
Baseline characteristics from the EHR dataset in the cohorts hospitalised before the COVID-19 pandemic (i.e.without COVID-19), and with high and low fibrinogen, before and after propensity score matching.

Table 10 :
Baseline characteristics from the EHR dataset in the cohorts hospitalised before the COVID-19 pandemic (i.e.without COVID-19), and with high and low D-dimer, before and after propensity score matching.