Proinsuliiis and their correlation with IGF and IGFBPs study

Characteristics of the study population are shown in Table 5.25. After adjustment for age, sex and BMI, IGFBP-1 levels correlated significantly with those of HDL-cholesterol, insulin, intact proinsulin, des 31, 32 proinsulin, insulin sensitivity, jS-cell function and PAI-1 activity, but not with fasting plasma glucose, blood pressure or LDL-cholesterol (Table 5.26).

To further assess the association of IGFBP-1 levels and the risk factors studied, multiple regression analysis were performed adjusting for age, BMI and insulin levels (Table 5.27). The association of IGFBP-1

concentrations with those of both HDL-cholesterol and triglycerides, and insulin precursors, remained strong.

The correlation of the other IGFs studied (IGF-1, IGF-2) and IGFBP-3, with the cardiovascular risk factors studied were not significant, either in univariate analysis or after adjusting for age, BMI and sex.

5.3.7 Proinsulins in insulinomas - study 7

Clinical characteristics of the subjects with insulinoma are shown in Table 5.28. The proinsulin-like molecules as a proportion of total

immunoreactive insulin ranged from 7.5 to 89.8%. Intact proinsulin was the predominant insulin precursor in 9 of the subjects and in 11 of the subjects the des 31, 32 species was predominant. While des 64, 65 proinsulin was detected in all the samples assayed it was not the

predominant precursor in any of the samples studied (Table 5.29). Des 64,65 proinsulin comprised 4% (SD 2%) (range 1% - 11%) of IRI in these

patients.

5.4 Discussion

The seven different applications to which the assays for the insulin precursors were applied allow some speculation of a possible role for the proinsulins in the various conditions.

The studies in non diabetic subjects (studies lA and IB) examined the association of proinsulins with other known cardiovascular risk factors, and secondly the extent to which these molecules contribute to the

hyperinsulinaemia seen in Asian and Caucasian non diabetic subjects. The first study (study lA) showed, for the first time, in a large homogeneous population comprising only Caucasians with normal glucose tolerance, that the levels of proinsulin-like molecules correlated with some insulin

resistance syndrome variables. After correcting for age and BMI

concentrations of des 31, 32 proinsulin were more strongly related to levels of triglycerides, HDL-cholesterol and LDL-cholesterol, and intact

proinsulin to fibrinogen, than were levels of insulin per se. In these subjects, on adjusting for BMI and age, no significant relations were apparent between the proinsulins and blood pressure, perhaps these being present only in the glucose intolerant state (Haffner et al, 1993). The finding of strong and independent relationships between levels of the proinsulins and cardiovascular risk factors in this low risk population raises questions regarding the likelihood of cause and effect.

The second study (study IB) of non diabetic Asian and Caucasian subjects, with normal and impaired glucose tolerance, showed that the Asian non diabetic subjects were hyperinsulinaemic after a glucose load compared with matched white subjects and this difference could not be attributed to age, sex, BMI or central obesity. Although Asian subjects with NGT showed higher absolute levels of intact proinsulin, the percentage of proinsulin-like molecules was similar in Asian and white subjects with IGT

and NGT. Also, in the basal state, the sum of all insulin-like molecules was similar in the two ethnic groups. The subject groups differed in that the Asian non diabetic subjects were more centrally obese and showed an exaggerated insulin response to a glucose load, perhaps due to enhanced pancreatic /3-cell sensitivity or a decreased hepatic extraction of insulin.

A study of insulin resistance, insulin-like molecules and blood pressure in subjects with non-insulin-dependent diabetes (study 2) confirmed high concentrations of the proinsulin-like molecules in NIDDM (Ward et al, 1987). No relation was seen between concentrations of intact and des 31, 32 proinsulin and any measures of blood pressure, clinic or 24-h

ambulatory, but they correlated strongly with glucose MCR, suggesting that the association of proinsulins with hypertension may be their common link with insulin resistance.

The effect of hypoglycaemic therapy on proinsulin-like molecules was investigated in two studies (studies 3A and 3B), one comparing metformin and sulphonylurea, and the second insulin and sulphonylurea. The first study (study 3A) showed that in subjects with NIDDM, treatment with metformin was associated with a reduction in levels of intact and des 31, 32 proinsulin, while glibenclamide treatment was not associated with such changes, although both therapies produced reduction in fasting plasma glucose concentrations. The ratio of concentrations of proinsulin-like molecules were also significantly reduced by metformin treatment, while it remained unchanged on glibenclamide. The precise mechanism of the fall in the concentrations of the proinsulin-like molecules with metformin therapy is unclear. However data suggests that a possible explanation may be that the action of metformin includes reducing insulin resistance (Nagi et al, 1996), and concentrations of proinsulin-like molecules may be a marker for such insulin resistance.

investigated in 20 subjects with NIDDM (study 3B) in a cross-over fashion with a 4-week washout period in between the two treatments. The

improvements in glycaemia and insulin sensitivity did not differ between the two treatments, but sulphonylurea therapy resulted in higher circulating levels of proinsulins and PAI-1 activity and antigen compared to the

baseline, and PAI-1 antigen levels were higher when compared to insulin therapy. The long term consequences of the increased proinsulins and PAI-

1 resulting from sulphonylurea treatment are at present unclear.

Fasting serum levels of intact and des 31, 32 proinsulin were increased 3 to

10-fold in the diabetic cirrhotic subjects in comparison to the non diabetic cirrhotic or control subjects (study 4). The higher levels of the propeptides seen in the cirrhotic diabetic subjects were similar to those reported in obese IGT subjects and in patients with NIDDM without liver disease (Yudkin, 1993; Levy et al, 1993). Circulating levels of the proinsulins are a product of their relative rates of secretion and clearance. In normal subjects proinsulin secretion is about 2 to 5 % of that of insulin (Horwitz et al, 1975). Intact proinsulin secretion is slightly greater than that of des 31, 32 proinsulin but the clearance of des 31, 32 proinsulin is higher thzm that of intact (Tillil et al, 1990; Given et al, 1985). The metabolic clearance rate of proinsulins is only about 20 - 25 % of that of insulin (Katz and Rubenstein, 1973; Tillil et al, 1990). The liver is the major site of insulin clearance accounting for about 70% (Ferrannini et al, 1983). The

kidneys, in contrast, are much more important in proinsulin clearance. In cirrhotic subjects renal function is normal with liver function being

impaired. Thus absolute levels of proinsulins were raised in both diabetic and non diabetic cirrhotic subjects, with the diabetic cirrhotic subjects having the highest levels. But because of reduced insulin clearance in cirrhosis the percentage of the proinsulin-like molecules to the total immunoreactive insulin is not raised (< 15%). Therefore the

hyperinsulinaemia in cirrhosis is not due to elevated levels of proinsulin­ like molecules.

The relationship between PAI-1 and the proinsulins were examined in two studies (studies 5 A and 5B). The first study (study 5A) looked at this interaction in Asian and Caucasian subjects with NIDDM, IGT and NGT. After adjusting for age and BMI, PAI-1 was more strongly related to the proinsulins than to insulin itself. PAI-1 activity was similar in Asians and Caucasians, but elevated in diet treated NIDDM subjects. While the elevated levels of triglycerides and the insulin-like molecules may contribute to the elevation in the PAI-1 activity in NIDDM, they do not fully explain it.

Similarly in diabetic and non diabetic survivors of myocardial infeu’ction (study 5B) the concentrations of proinsulin-like molecules and serum triglycerides appeared to be the major determinants of PAI-1 activity, with closer correlations than either with insulin sensitivity or insulin levels. Again a causal relationship appears unlikely, and the relationship may be indicative of an association via a common antecedent.

The study of the relationship between the IGFs, their binding proteins and the proinsulins (study 6) showed independent and significant correlations between levels of IGFBP-1 and levels of lipoproteins, insulin and

proinsulin-like molecules and insulin resistance, but no such relationships with IGF-1, IGF-2 or IGFBP-3. Both intact and des 31, 32 proinsulin showed significant negative correlations with IGFBP-1 after correcting for insulin levels or insulin resistance.

One explanation for this relationship may be that the insulin precursors with their high sequence homology to IGF-1 and IGF-2 may directly bind to IGFBP-1 and thus give rise to the correlation demonstrated.

Alternatively these findings may indicate that circulating insulin propeptides are a better index of portal insulin delivery than peripheral insulin

concentrations, on account of the longer half-life of the precursors in the circulation (Sodoyez-Goffaux et al, 1988). Although concentrations of the

proinsulins are elevated in NIDDM their activity at the insulin receptor is <10% (Peavy et al, 1985). Thus, the strong inverse relationship between concentrations of IGFBP-1 and the proinsulins may reflect the negative regulation of hepatic IGFBP-1 by insulin or insulin resistance.

The last study investigated the relative contribution of each of the

proinsulin-like molecules, intact, des 31, 32 and des 64, 65 proinsulin, to the hyperinsulinaemia seen in subjects with insulinomas (study 7). The constitutive release of insulin in these subjects would suggest that the total concentration of insulin-like molecules would consist of a higher proportion of propeptides, even when compared with other hyperinsulinaemic states. This does appear to be the case, with 25 % of the patients studied having proportions of proinsulin-like molecules of 50% or over of total

immunoreactive insulin. Des 64, 65 proinsulin was detected in all the samples analyzed, but surprisingly, none of the insulinomas secreted this propeptide species predominantly. One previous report suggested that there is heterogeneity in the proinsulins secreted in subjects with NIDDM, with some subjects having a predominance of des 64, 65 proinsulin and others secreting des 31, 32 proinsulin (Linde et al, 1991). The present results, however, support the findings that proinsulin to insulin processing occurs predominantly via the des 31, 32 proinsulin branch, even in insulinoma.

In conclusion, all the studies show that concentrations of the propeptides are raised in NIDDM, they appear to show statistical interaction with a range of cardiovascular risk factors, including lipids, fibrinogen, PAI-1 and IGFBP-1. Of course, statistical correlation is not proof of causation, so that proinsulins cannot necessarily be implicated directly in determining abnormal levels of risk factors. The cross-reactivity of these assays with non-specific assays for insulin was previously thought to account for the hyperinsulinaemia seen in various conditions. However, these studies, using specific assays for insulin, intact proinsulin and des 31, 32

proinsulin, still demonstrates the presence of hyperinsulinaemia. Also, the

study of these molecules in insulinoma samples appear to suggest that the des 64, 65 proinsulin species may be of very little physiological