1. Early and sustained acute falls in total serum cholesterol are associated with critical illness mortality in the setting of tight glycaemic control Paul Dark Infection, Injury and Inflammation Research Group Hope Hospital Intensive Care Unit The University of Manchester, UK

2. Motivation Wellcome Trust Host-Pathogen Consortium Steve Kemp and Andy Brass T. congolense infection in mice (Sepsis/CARS progression) Quantitative trait loci (QTLs) underlying resistance have been determined and various congenic mice constructed (C57BL/6 resistant; BALB/c susceptible)

3. Motivation Micro array: Mice tolerant to infection have a higher expression of important components of cholesterol biosynthesis (HMGCoA reductase) and low expression of srb-1, abcg-1, and apoA-1 genes which are involved in cholesterol transport compared with the susceptible strains. Kierstein S, (2007) Genes and Immunity, In press RXR isoforms/Vit D3/retinoic acid (? Role in M1 and M2 phenotype)

4. Motivation Phenotype (unpublished): Mice susceptible (BALB/c) to parasite infection have markedly decreased circulating cholesterol and shorter survival times, higher mortality Mice pre-fed on the high fat diet increased significantly in weight had higher plasma cholesterol and triglyceride levels, and lower parasitaemia, severe anaemia than the low fat-fed controls

5. Introduction Serum cholesterol is known to be associated with survival in critical illness, but this has not been verified in an environment of tight glycaemic control (a potential confounder) The effects of changes in cholesterol levels have not been studied

6. Introduction Study question: Are serum cholesterol levels and their early changes independently predictive of mortality at 28 days from admission to intensive care, where tight glycaemic control is normal practice?

7. Methods Design: Prospective case cohort study Setting: University Hospital level 3 Adult ICU (mixed trauma/surgical/medical) Population: All critically ill patients (01-April 2006 – 31-January 2007) Interventions: Addition of lipid testing to routine daily blood analysis without ICU staff intervention - data reflect usual ICU practice Total cholesterol, triglycerides, HDL cholesterol (mmol/l) Roche Diagnostics Modular Analyser Outcome measure : Alive or dead at 28 days following initial ICU admission

8. Methods Tight glycaemic control: -nurse delivered (high ratio of nurse/ patient) -web-based support (insulin dose calculator) -continuous feed (80-90% enteral delivery) -upper limit of control 7.1 mmol/l -extremely low rates of hypoglycaemia Ref: Thomas A. el al. Anaesthesia 2005 60:1093 - 1100

9. Results Population characteristics: Case cohort size = 607 Male = 57% Median Age = 56 years Dead within 28 days = 33% APACHE II

10. Results Population characteristics: Case cohort size = 607 Male = 57% Median Age = 56 years Dead within 28 days = 33% APACHE II = median of 16 Severe sepsis on admission = 23%

11. Results Population characteristics: Case cohort size = 607 Male = 57% Dead within 28 days = 33% APACHE II = median of 16 Severe sepsis on admission = 23% ICU length of stay

12. Results Glycaemic control

13. Results Triglyceride responses

14. Results Total cholesterol responses

15. Results HDL cholesterol responses

16. Results Non-HDL cholesterol responses

17. Results ICU Day 1 analysis: Patients who will survive to 28 days have a significantly higher total serum cholesterol on admission (95% CI difference = 0.04 to 0.58, P = 0.02) Patients with severe sepsis have significantly lower total cholesterol on admission (95% CI for difference = 0.7 to 1.3, P < 0.0001)

18. Results ICU Day 1 analysis: Patients who will survive to 28 days have a significantly higher total serum cholesterol on admission (95% CI difference = 0.04 to 0.58, P = 0.02) Patients with severe sepsis have significantly lower total cholesterol on admission (95% CI for difference = 0.7 to 1.3, P < 0.0001) Logistic regression model Day 1 cholesterol predicts 28 day mortality OR = 0.83, 95%CI 0.69-0.87 (p=0.03) Independent of gender but not age, severe sepsis or APACHE II score

19. Results Analysis of early total cholesterol responses Survived = mean increase in total cholesterol on second day (0.13 mmol/l) Died = mean decrease in total cholesterol on second day (0.10 mmol/l) P=0.001, 95% CI for difference = 0.10 to 0.37

20. Results Analysis of early HDL cholesterol responses

21. Results Analysis of early non-HDL cholesterol responses

22. Results Analysis of changes over first 72 hours: Logistic regression model A fall in cholesterol between days 2 and 3 of admission to ICU predicts death within 28 days OR 0.29 95%CI 0.14 - 0.60 (p<0.001) Independent of APACHE II score, severe sepsis or baseline cholesterol These patterns are not seen for HDL cholesterol or triglycerides The prognostic value of total cholesterol change is independent of glucose

23. Conclusions While establishing tight glycaemia: Total cholesterol is associated with subsequent survival to 28 days Prognostic value is linked to severe sepsis and APACHE II Not seen in HDL-cholesterol

24. Conclusions While establishing tight glycaemia: Total cholesterol is associated with subsequent survival to 28 days Prognostic value is linked to severe sepsis and APACHE II Not seen in HDL-cholesterol Established tight glycaemia: Switching to cholesterol recovery (Day 2-3) is associated with survival Prognostic value independent of severe sepsis, APACHE II and baseline cholesterol Not seen in HDL-cholesterol

25. Models of tryp. Livestock/human observation/studies Models of critical illness Patient observation/studies

26. Infection, Injury and Inflammation Research Group Prof. Geoff Warhurst Population and Endocrine Medicine Dr. John New Dr. Martin Gibson Clinical Chemistry Dr. Aram Rudenski Northwest Institute of Biomedical and Health Informatics Dr. Iain Buchan Dr. Pat Baker Wellcome Trust Host-Pathogen Consortium Prof. Steve Kemp Prof. Andy Brass