File S1 - Thresholds of Oxidative Stress in Newly Diagnosed Diabetic Patients on Intensive Glucose-Control Therapy

File includes Figures S1–S29 and Tables S1–S8. Figure S1: Cluster analysis of GSH values pulled together from non-diabetics and diabetics 0 and 8 weeks, age group below 40. •: non-diabetics 0 and 8 weeks (n = 72), ○: diabetics 0 week (n = 11), △: diabetics 8 weeks (n = 11). Three clusters emerged from the cluster analysis. Unlike the cluster analysis for the age group above 40 as shown in the main text, Fig. 2, below 40 GSH values do not show separation within diabetics groups 0 and 8 weeks. However, the non-diabetic below-40 age group is separated into two clusters, which shows apparent within-group age dependence on GSH levels. Figures S2–S9: Serial changes in plasma insulin and glucose for diabetic cases 1–54. ▴: 0 week, △: 4 weeks and •: 8 weeks. Figure S10: Average change in plasma glucose levels in diabetics kept on the anti-diabetic treatment for 8 weeks (n = 46). Mean and standard deviation values of plasma glucose corresponding to 0, 4 and 8 weeks are 10.7±3.3, 8.3±2.3 and 7.6±1.7, respectively. Paired t-test of mean change in plasma glucose at 0 and 8 weeks shows statistical significance, with p-value <0.05 at a 95% confidence interval. Figure S11: Average change in plasma insulin levels in diabetics kept on the anti-diabetic treatment for 8 weeks (n = 46). Mean and standard deviation values of plasma insulin corresponding to 0, 4 and 8 weeks are 11.6±8.2, 11.6±7.3 and 12.0±8.7, respectively. Though, there is slight increase in insulin secretion over 8 weeks, paired t-test of mean change in plasma insulin at 0 week and 8 weeks is not statistically significant, with p-value = 0.7 at a 95% confidence interval. Figure S12: HOMA2-IR against Glucose for non-diabetics and diabetics at 0 and 8 weeks. The bold line indicates serially observed change in the diabetics while the dotted line shows is a projection that assumes that if diabetics were to continue on the therapy for longer time period the asymptotic values of HOMA2-IR may lie close to the non-diabetic numbers. Figure S13: HOMA2-%B against GSH for non-diabetics and diabetics at 0 and 8 weeks. As in the previous figure, the bold line indicates serially observed change in the diabetics and the dotted line shows is a projection that assumes that if diabetics were to continue on the therapy for longer time period the asymptotic values of HOMA2-%B may lie close to the non-diabetic numbers. Figure S14: Linear regression of GSH against age in non-diabetics (n = 48). GSH levels are affected due to aging in non-diabetics. The equation for this regression line is GSH  = 1354.5–14.3×age, where p-values for the intercept and slope being <0.05 and 0.0002, respectively, at a 95% confidence interval. BMI doesn't contribute to GSH levels significantly (Data not shown, p-value for the slope of −5.24 being 0.73 at a 95% confidence interval). Figures S15–S23: Individual sigmoid fits for diabetic cases 1–54. In each case the diabetic patient's glucose and GSH pair at 0 week (□), 4 weeks (○) and 8 weeks (△) are shown alongside glucose-GSH pair taken from non-diabetic subjects from their first visit (▴) using regression fit. The pathophysiological parameters v, k and Gtot estimated from a fit are displayed in its panel. Figure S24: Individual response curves for diabetic patients above age 40 obtained using the minimal model are shown (thin gray lines, n = 29 out of 38) along with population-averaged curve (black bold line). The population-averaged curve has a threshold (black dot) at glucose  = 7.5 mmol/L and GSH approximately 347; Gtot = 695 and k = 43. An inflection regime is marked in red (width approximately one fourth of v, 1.87 mmol/L) is marked in red. The ADA impaired fasting glucose (IFG) range, 5.5–6.9 mmol/L and WHO IFG range 6–6.9 mmol/L is overlaid for the reference. The GSH band at 220–480 is the recovery phase for treated diabetics as shown in the main text, Figure 2. It is interesting to note that IFG occupies upper portion of the red curve, and 8-weeks patients lie in the lower portion of the red curve. Figure S25: Individual response curves for diabetic patients below age 40 obtained using the minimal model are shown (thin gray lines, n = 5 out of 11) along with population-averaged curve (black bold line). The population-averaged curve has a threshold (black dot) at glucose  = 7.4 mmol/L and GSH approximately 462; Gtot = 924 and k = 48.7. An inflection regime is marked in red (width approximately one fourth of v, 1.85 mmol/L) is marked in red. The ADA impaired fasting glucose (IFG) range, 5.5–6.9 mmol/L and WHO IFG range 6–6.9 mmol/L is overlaid for the reference. Unlike the above 40 group, cluster analysis does not show GSH separation for diabetics recovery as shown in the figure S1. Nonetheless, the IFG band lies in the sensitive upper portion of the red curve. Figure S26: A comparison of GSH values of non-diabetic subjects and theoretical predictions of GSH values of diabetic patients at glucose were 5.2 mmol/L. This plot shows the natural variability in GSH at low glucose, in non-diabetics and diabetics. Figure S27: Distributions of v, k and Gtot in the diabetics above age 40. Mean and standard deviation values for v, k and Gtot are 7.5±1.1, 43.0±40.0 and 695±166, respectively. Figure S28: Linear regression between fasting glucose and HbA1c. Fasting glucose and Hb1Ac values were taken from •: Non-diabetic; °: diabetic 0 week; □: diabetic 4 weeks; △: diabetic 8 weeks. This equation is used to convert HbA1c into a glucose value for model fitting. Figure S29: Distributions of the parameters v, k and Gtot for the samples cases 15 and 13. Table S1: Summary of anthropomorphic characteristics: Gender, age and BMI of non-diabetics (n = 48) and diabetics (n = 49) used in the data analysis. Table S2: Summary of anti-diabetic drug treatment given to 48 diabetic subjects over the period of 8 weeks. Out of 48 diabetics, 58% received DPP-4 inhibitor or gliptin treatment, 21% received biguanide drug treatment and remaining 21% received combination of biguanides and sulphonamides drug treatment. Table S3: Multiple linear regression of 0-week GSH with Age and BMI, in non-diabetics (n = 12) and diabetics (n = 38) above age 40. Both age and BMI are not significant predictors of GSH within non-diabetic and diabetic groups. Table S4: Multiple linear regression of 0-week GSH with age and BMI, in non-diabetics (n = 36) and diabetics (n = 11) below age 40. In both groups BMI is not significant predictor of GSH. However, age predicts GSH in non-diabetics, but not in diabetics. Table S5: Mean and standard deviation values corresponding to normal or log-normal probability density curves fitted to GSH and HbA1c levels of non-diabetics and diabetics shown in the Figure 1 in the main text. Table S6: A hierarchical cluster analysis performed on GSH values of non-diabetics (n = 23), diabetics at 0 week (n = 38) and diabetics at 8 weeks (n = 38) showed 3 clusters emerging from the data. For example, cluster 1 comprises of 3 diabetics at 0 week, 27 diabetics from 8 weeks and 5 non-diabetic. Based on this information we could distinguish between diabetics, before and after treatment, and non-diabetics, as shown in the Figure 2 in the main text. Table S7: Mean and standard deviation values corresponding to normal or log-normal probability density curves fitted to GSH and HbA1c levels of non-diabetics and diabetics below age 40. Table S8: A hierarchical cluster analysis performed on GSH values of non-diabetics (n = 72) 0 and 8 weeks together, diabetics at 0 week (n = 11) and diabetics at 8 weeks (n = 11) showed 3 clusters emerging from the data.
(PDF)