Lipid composition of plasma lipoproteins in treated diabetics.
In order to assess the disturbances of lipid metabolism in clinical practice, it has been customary to determine plasma total cholesterol and triglycerides, and plasma lipoprotein fractions by electrophoresis, and to define hyperlipoproteinemia in terms of their levels. Elsewhere in this journal, Retnam et al have reported a study of plasma lipids and lipoproteins in diabetics on different modes of treatment. However, it has recently become apparent that the lipid composition of the individual lipoprotein fractions (cholesterol and triglyceride content of HDL, LDL and VLDL) may be more important as parameters of abnormal lipid metabolism in assessing the risk of ischemic heart disease., ,  Hence, a study of the cholesterol and triglyceride composition of HDL, LDL and VLDL fractions of plasma was undertaken in normolipemic and hyperlipemic diabetics on different modes of treatment, attending the Department of Endocrinology, K.E.M. Hospital. For the purpose of this study, `normolipemia' was defined as normal serum levels of total cholesterol and triglycerides, and normal lipoprotein electrophoretogram.
The purpose of this study was dual (a) to delineate the abnormalities in the lipid composition of lipoprotein fractions in diabetics, with reference to the mode of treatment, control or otherwise of hyperglycemia, and the presence or absence of hyperlipoproteinemia (HLP) by the criteria described earlier; and (b) to decide the optimum mode of treatment which would be associated with `near cholesterol, LDL-triglyceride and VLDL-triglyceride.
The study group and the controls were the same as those described in our earlier paper. The analyses described below were carried on the same plasma samples used in that study. All analyses were started on the same day as the blood collection. Total cholesterol was estimated by Abell's method; total triglycerides by Van Handel and Zilversmit's method. Lipoprotein fractionation was carried out according to the method of Onongbu and Lewis. Cholesterol and triglyceride (TG) content of VLDL and HDL were actually measured and those of LDL were calculated by subtraction from total plasma cholesterol and TG values. All parameters (HDL cholesterol, LDL cholesterol, VLDL cholesterol, HDL triglyceride, LDL triglyceride and VLDL triglyceride) were expressed in mg per 100 ml of plasma. The data were analysed by two tailed, unpaired, Student's `t' test. In this paper, the term `near normal' is used in the sense that the value is not statistically different from that in the normal controls.
Each diabetic patient in this study was classified in two ways: (a) hyperlipoproteinemic or normolipemic, using the criteria described in our earlier paper; and (b) abnormal or normal using (mean + 2 SD) as the cut off point in case of HDL cholesterol, and (mean + 2 SD) as the cut off paint in case of LDL cholesterol, LDL-TG and VLDL-TG. Sensitivity, specificity, positive accuracy and negative accuracy of the package of older tests (total plasma cholesterol and TG, and lipoprotein electrophoretogram) were calculated considering HDL cholesterol, LDL cholesterol, LDL-TG and VLDL-TG as the final arbiters of lipid abnormalities.
The demographic data of the study population and the controls were the same as that described in our earlier paper.
[Table - 1] shows the means and standard deviations of different parameters in normal controls, controlled diabetics, and uncontrolled diabetics. As seen from the table, LDL cholesterol and HDL-TG in controlled diabetics were `near normal'. The remaining parameters in the controlled diabetics and all the parameters in the uncontrolled diabetics were significantly different from those in the normal controls. In particular, the HDL cholesterol was significantly lower in both uncontrolled and controlled diabetics compared to the normal controls.
[Table - 2] shows the cholesterol content of the various lipoprotein fractions in normal controls, normolipemic diabetics and diabetics with different types of HLP. The division of the diabetics into normolipemic and hyperlipoproteinemic is the same as in our earlier paper. LDL cholesterol was `near normal' in normolipemic diabetics. It was significantly higher than in normal controls in all groups of hyperlipemic diabetics. HDL cholesterol was significantly lower than in normal controls in all groups.
[Table - 3] shows the TG content of various lipoprotein fractions in normolipemic diabetics and diabetics with different types of HLP. VLDL-TG was `near normal' in the normolipemic diabetics but was significantly higher than in normal controls in diabetics with all varieties of HLP. LDL-TG on the other hand was significantly higher than in the normal controls in all diabetics groups.
[Table - 4] shows the values of the cholesterol content of different lipoprotein fractions in normal controls, controlled diabetics, uncontrolled diabetics and in the combined (controlled + uncontrolled) group of diabetics on different treatment regimes. As in our earlier paper, this table indicates only those comparisons where the results are not statistically significant even at the 5% level (i.e. p 0.05) i.e. those that are `near normal'. The other comparisons gave results which were statistically significant at least at the 5% level. This method has been followed because one of our purposes was to find out which mode of treatment gave results close to those observed in normal controls. As the table shows, all parameters showed `near normal' results in patients on insulin whereas HDL cholesterol and LDL cholesterol were `near normal' in patients well-controlled on diet. Patients on a sulphonylurea, controlled as well as uncontrolled, showed significant differences from normal controls in the cholesterol content of all lipoproteins. Well controlled patients on combined sulphonylurea + phenformin therapy had `near normal' LDL cholesterol but their HDL cholesterol was significantly lower than that in the normal controls.
[Table - 5] shows the TG content of the different lipoprotein fractions in normal controls and in different treatment groups. As in [Table - 4] only those comparisons are indicated where the results are not statistically significant even at the 5% level (p > 0.05). As the table shows, the LDL-TG and VLDL-TG were `near normal' in patients on insulin. In all other groups, they were significantly higher than in normal controls.
[Table - 6] shows the cholesterol and TG content of the different lipoprotein fractions in normolipemic diabetics on different modes of treatment. Indicated in this table are only those comparisons where the difference from the normal controls was statistically significant. This is in contrast to the practice followed in [Table - 4] and [Table - 5]. We would like to focus the attention of the reader on HDL cholesterol, LDL cholesterol, LDL-TG and VLDL-TG. HDL cholesterol levels were significantly lower than in normal controls in all groups except those on either insulin or diet. On the other hand the LDL cholesterol levels were `near normal' in all groups. Again, the LDL-TG levels were significantly higher than in normal controls in all groups except those on diet or insulin. But, VLDL-TG was `near normal' in all groups.
[Table - 7] shows the sensitivity, specificity, positive accuracy and negative accuracy of the package of older, tests in terms of individual lipid content of certain lipoprotein fractions. The table suggests that sensitivity of the older tests is high (90 for more) in the case of LDL cholesterol and VLDL-TG. It is lower in the case of LDL-TG, and is particularly low in the case of HDL cholesterol. Negative accuracy (which answers the question, "If the test is negative, is the actual condition likely to be negative?" in individual patients rather than, unlike sensitivity, in populations) gives similar results-it is lowest in the case of HDL cholesterol.
This study substantiates the impression of the previous paper that plasma lipid values closest to those in normal controls are seen in patients treated with insulin but not in those on sulfonylurea. Further, as a group, the diabetics in this study (controlled as well as uncontrolled) had significantly low HDL cholesterol levels [Table - 1] this was true even of those who were normolipemic [Table - 2]. However, HDL cholesterol levels were `near normal' in normolipemic diabetics treated by diet or insulin [Table - 6]. Further, HDL cholesterol levels were `near normal' in diabetics on insulin irrespective of the control of hyperglycemia [Table - 4] On the other hand, patients on sulfonylurea had significantly low HDL cholesterol values irrespective of the control of hyperglycemia [Table - 4] irrespective of combination of sulfonylurea with phenformin or with insulin [Table - 4] and irrespective of the fact that they were normolipemic [Table - 6]. LDL cholesterol was `near normal' in normolipemic diabetics on all modes of treatment [Table - 6] Without reference to whether the patients were normolipemic or hyperlipoproteinemic, LDL cholesterol levels, were `near normal' in patients treated with insulin, and in those well controlled with diet, with sulphonylurea + phenformin combination or with sulphonylurea + insulin combination [Table - 4] But LDL cholesterol was significantly higher than in normal controls in patients on sulfonylurea alone, irrespective of the control of hyperglycemia [Table - 4].
As a group, the diabetics in this study (controlled as well as uncontrolled) had significantly elevated levels of VLDL-TG and LDL-TG [Table - 1]. However, VLDL-TG was `near normal' in those who were normolipemic [Table - 3] and [Table - 6] irrespective of the mode of treatment [Table - 6]. But, LDL-TG levels were significantly higher than in normal controls even in normolipemic patients [Table - 3] and [Table - 6] except in those on insulin and on diet [Table - 6]. On the other hand, all groups of patients on sulfonylurea had significantly elevated VLDL-TG and LDL-TG levels irrespective the control of hyperglycemia
[Table - 5] and irrespective of whether or not they received additional phenformin or insulin [Table - 5] But, in normolipemic diabetics on sulfonylurea, VLDL-TG levels, but not LDL-TG levels, were `near normal' [Table - 6]
An important point that emerges from this study is that even the `normolipemic' patients (except those on diet or on insulin) had abnormalities in the lipid composition of certain plasma lipoprotein fractions, viz. depressed HDL cholesterol and elevated LDL-TG levels. This is partly a reflection of the low `negative accuracy' of `normolipemia' for these two parameters as mentioned earlier [Table - 7]. Thus, it would appear appropriate to include normal plasma HDL cholesterol level as an additional, necessary criterion of `normolipemia'. This, in fact, is the current practice in the literature.
Abnormalities in the lipid composition of plasma lipoprotein fractions (elevated LDL-TG and low HDL cholesterol) in diabetes were first reported by Schonfeld et al. Howard et al extended this work and also reported elevated plasma levels of LDL cholesterol and VLDL-TG in diabetic Pima Indians; they further found that these abnormalities correlated with abnormal two hour post glucose blood sugar levels. Similarly, Lopes-Virella and Cowell found a negative correlation between plasma HDL cholesterol levels and fasting plasma blood glucose levels. On the other hand, Ballantyne et al could not find any difference in the lipid composition between diabetics and nondiabetics.
The beneficial effect of insulin therapy on the lipid composition of plasma lipoprotein fractions has been documented by many workers. For example, Bennion and Grundy as well as Michael et a1 reported that HDL cholesterol levels rose concurrently with the control of hyperglycemia with insulin. Nikkila and Hormila observed that insulin-treated diabetics had, in fact, higher HDL cholesterol levels and lower VLDL-TG levels than normal controls. In a review of continuous, subcutaneous, insulin infusion (CSII), Raskin has listed a fall in LDL cholesterol and a rise in HDL cholesterol as two of the metabolic effects of such therapy. Finally, Brown and Ginsberg have recently stated that "The universal observation has been that improved management with insulin leads to elevation of HDL, producing normal or even high values".
On the other hand, the effects of treatment with sulfonylurea on the lipid composition of plasma lipoprotein fractions have been controversial. For example, low HDL cholesterol levels in patients treated with sulfonylurea were reported by Bar-On et a1 and by Calvert et al. But, Paisey et al reported similar increases in HDL cholesterol levels in diabetics treated with either chlorpropamide or insulin, at the end of 12 months of treatment. The same workers observed an initial fall in VLDL-TG in both insulin treated and chlorpropamide treated diabetics, though this reduction was maintained at the end of 12 months only in insulin treated patients.
The data on lipid composition of 'normolipemic' diabetics is scanty. Koschinsky et al have reported decreased HDL cholesterol levels in normolipemic diabetics on sulfonylurea treatment.
Thus, it will be seen that a consensus is emerging in the literature that insulin treated patients tend to have fewer abnormalities of the lipid composition of lipoproteins than those treated with sulphonylurea. Further, many workers, , , ,  feel that well controlled patients (on insulin) have better lipid profiles than those poorly controlled. The present study generally supports these contentions.
These observations could be of great relevance to the future treatment of diabetes with purer forms of insulin and newer insulin delivery systems such as portable insulin pumps. However, their relevance to the Indian population, at the present state of art and at the present cost and availability of insulin in India, is difficult to assess.
We thank the Dean, Seth G. S. Medical College and K.E.M. Hospital for allowing us to publish this hospital data.