The pilot cell culture experiments were done in duplicates (n=2) hence statistical analysis was not performed on these very small numbers. Analysis is presented in graphs as the mean of each condition to appreciate the dose response curves in the different basal conditions following stimuli with different doses of the same cytokine. This has been done for each of the cell lines used. This informed the decision to repeat the experiments of interest for statistical evaluation.
72 2.7.4 Statistical analysis for HK2 and Podocyte cell culture experiments
Subsequent cell culture experiments were repeated 3 times with each experiment having the same condition repeated in 6 wells (n=6) and hence analysed with non-parametric statistical analysis. Kruskal-Wallis and post analysis Dunn’s multiple comparison test allowed comparison of selected control data, i.e. Normal glucose compared with its control glycated albumin and Mannitol with High glucose. See table 2.6 for cell culture controls. In addition, the effect of cytokine stimulation (20ng/ml) within a certain condition was detected by the difference in levels from the condition without stimulation (0ng/ml). This change is analysed with Kruskal-Wallis and Dunns post analysis multiple comparisons with e.g. normal glucose 20ng/ml with normal glucose 0ng/ml. The effect of cytokine and fibronectin levels can then be determined. Analysis was performed on 0ng/ml and 20ng/ml in view of time limitations preventing ELISA analysis of 10ng/ml.
Table 2.6 Cell culture controls
Cell culture diabetic condition Control used for statistical analysis Glycated Albumin Normal glucose
High glucose Mannitol
The experiments were repeated three times, thus each condition had a total of 18 wells (n=18) over all three experiments. The results of these wells were not pooled together as this would introduce bias to the results, therefore, each experiment with 6 wells per condition was analysed (n=6) and the results reviewed to establish whether the findings were consistent. For ease of understanding the results of a single experiment have been shown throughout Chapter 4, for each cell type, conditions and cytokine stimulations.
73 CHAPTER 3.0 – RESULTS – PROSPECTIVE CLINICAL COHORT
This chapter will determine what the demographics are for the clinical cohort at baseline, 18 months and >3 year time points. It will explore the relationship of baseline urinary and plasma MIF with demographics and outcomes; GFR, ACR and UPCR in DM and Non-DM groups. The baseline relationship between MCP-1 and CCL18 with outcomes; GFR, ACR and UPCR will be analysed to allow direct comparison of these between the DM and Non-DM groups. The latter data has previously been analysed in a different manner by Dr Qureshi MD 2012 who began the database for this cohort and gave me permission to use data from the original database. Clinical correlations between MCP-1 and CCL18 have been previously analysed by Dr Qureshi. I collected further data points for all patients in the cohort over time and amended the database to allow for the new data. I performed MIF ELISAs on all the available baseline urine and plasma samples from the original cohort.
The relationship between baseline urinary and plasma MIF, MCP-1 and CCL18 will be analysed with GFR, ACR and UPCR at the different time points. The relationship between GFR and ACR or UPCR in the different groups will be analysed at baseline and at >3 years’ time point to determine whether this changes over time. Comparison will be made between those in DM or Non-DM groups. The cytokines relationship at >3 years with GFR, ACR and UPCR will also be analysed to determine whether these change over time. Comparison of the urinary and plasma cytokine measurements at baseline to cytokines at >3 years urinary and plasma samples will also be made. This will determine the direction of change and whether this is different in DM or Non-DM patients.
I performed a univariant analysis to determine the effects of the cytokines in rate of change in GFR, ACR or UPCR in DM and Non-DM groups, respectively. This looks to see whether baseline cytokines are related to the outcome measures above, over time. There are limitations in using univariant analysis as this does not consider changes due to different variables such as age, gender, BMI. Mixed models that account for these variables also have limitations as they do not allow identification of those in a group that may have faster deteriorating GFRs. The results of these analyses are presented herein with their limitations. The mixed model of statistics was performed by Dr Gordon (Imperial College Statistician) on the data points and samples I collected and analysed during this study. This will help determine whether a single test at baseline will be able to predict a worse outcome in the DM group.
74 Finally this chapter will look at the differences between plasma and serum cytokines in all the samples collected at >3 years. This will determine whether samples can be taken in the same tubes as other blood tests without affecting the levels of different cytokines present. In addition, the profile of the cytokines at baseline and at >3 years will be compared and whether being on RRT affects MIF, MCP-1 or CCL18 levels regardless of DM or Non-DM groups. The latter is a small analysis and requires interpretation with caution, as larger sample size would be required and the effects of RRT on cytokines are multifactorial.