King's College London

Research portal

Cortical cerebral blood flow in ageing: effects of haematocrit, sex, ethnicity and diabetes

Research output: Contribution to journalArticlepeer-review

Lorna A. Smith, Andrew Melbourne, David Owen, M. Jorge Cardoso, Carole H. Sudre, Therese Tillin, Magdalena Sokolska, David Atkinson, Nish Chaturvedi, Sebastien Ourselin, Alun D. Hughes, Frederik Barkhof, H. R. Jäger

Original languageEnglish
Pages (from-to)5549-5558
Number of pages10
JournalEuropean Radiology
Issue number10
Early online date18 Mar 2019
Accepted/In press7 Jan 2019
E-pub ahead of print18 Mar 2019
Published1 Oct 2019


King's Authors


OBJECTIVES: Cerebral blood flow (CBF) estimates from arterial spin labelling (ASL) show unexplained variability in older populations. We studied the impact of variation of haematocrit (Hct) on CBF estimates in a tri-ethnic elderly population.

MATERIALS AND METHODS: Approval for the study was obtained from the Fulham Research Ethics Committee and participants gave written informed consent. Pseudo-continuous arterial spin labelling was performed on 493 subjects (age 55-90) from a tri-ethnic community-based cohort recruited in London. CBF was estimated using a simplified Buxton equation, with and without correction for Hct measured from blood samples. Differences in perfusion were compared, stratified by sex, ethnicity and diabetes. Results of Student's t tests were reported with effect size.

RESULTS: Hct adjustment decreased CBF estimates in all categories except white European men. The decrease for women was 2.7 (3.0, 2.4) mL/100 g/min) (mean (95% confidence interval (CI)), p < 0.001 d = 0.38. The effect size differed by ethnicity with estimated mean perfusion in South Asian and African Caribbean women found to be lower by 3.0 (3.6, 2.5) mL/100 g/min, p < 0.001 d = 0.56 and 3.1 (3.6, 2.5) mL/100 g/min), p < 0.001 d = 0.48, respectively. Estimates of perfusion in subjects with diabetes decreased by 1.8 (2.3, 1.4) mL/100 g/min, p < 0.001 d = 0.23) following Hct correction. Correction for individual Hct altered sample frequency distributions of CBF values, especially in women of non-European ethnicity.

CONCLUSION: ASL-derived CBF values in women, non-European ethnicities and individuals with diabetes are overestimated if calculations are not appropriately adjusted for individual Hct.

KEY POINTS: • CBF quantification from ASL using a fixed Hct of 43.5%, as recommended in the ISMRM white paper, may lead to erroneous CBF estimations particularly in non-European and female subjects. • Individually measured Hct values improve the accuracy of CBF estimation and, if these are not available, an adjusted value according to gender, ethnicity or diabetes status should be considered. • Hct-corrected ASL could be potentially important for CBF threshold decision making in the fields of neurodegenerative disease and neuro-oncology.

Download statistics

No data available

View graph of relations

© 2020 King's College London | Strand | London WC2R 2LS | England | United Kingdom | Tel +44 (0)20 7836 5454