Plasma Prekallikrein: A Risk Marker for Hypertension and Nephropathy in Type 1 Diabetes

The study population was the North American DCCT/EDIC cohort, which comprised 1,325 type 1 diabetic patients from an original 1,441 DCCT subjects. The original DCCT cohort consisted of men and women ages 13–40 years with diabetes of 1–15 years’ duration at study entry (21) who were enrolled between 1983 and 1989. Half of the patient population was randomly assigned to conventional diabetes treatment and the other half was assigned to intensive diabetes treatment. In 1993, the DCCT study was stopped after an average follow-up time of 6.5 years, when intensive treatment was clearly shown to reduce the risks of retinopathy, nephropathy, and neuropathy (21). The patients were then invited to enroll in the EDIC, a multicenter, longitudinal observational study of the development of macrovascular complications and further progression of microvascular complications (22). At EDIC baseline in 1994, the average age of the DCCT/EDIC cohort was 35 years (range 19–50 years), with 54% of the cohort being male and the mean duration of diabetes being 12 ± 5 years.

In 1996, a collaborative project (Markers and Mechanisms for Macrovascular Disease in Type 1 Diabetes) was undertaken by the Medical University of South Carolina (MUSC) and EDIC. Its goal was to identify new markers, risk factors, and mechanisms for vascular disease in type 1 diabetic patients using plasma samples shipped directly from participating EDIC clinics to MUSC. In all, 25 of the 28 EDIC clinics participated, and between 1997 and 1999, fasting plasma samples were collected from subjects for KKS component measurements. The Institutional Review Boards of MUSC and all participating DCCT/EDIC clinics approved the study, and written informed consent was obtained from each patient participant.

On the approximate anniversary of enrolling in the DCCT, each EDIC subject had a standardized annual history and physical examination, including a detailed evaluation of overall health, diabetes management, occurrence of diabetic complications, development of new disease, and medications used. Annual evaluations also include HbA_1c, resting electrocardiograms, and arm blood pressure (BP) measurements. BP and HbA_1c measurements were done at the same time as blood sampling. BP was measured in the right arm using a mercury column sphygmomanometer while the patient was in the sitting position. Renal function was assessed every second year and included measurements of the urinary albumin excretion rate (AER) in a standardized 4-h collection (21). The DCCT has defined microalbuminuria as an AER of 40–300 mg/24 h; albuminuria, as an AER ≥300 mg/24 h; and normal albumin excretion, as an AER ≤40 mg/24 h. In all, ∼98% of all AER measurements in the EDIC subjects were carried out within 1 year of blood sampling.

Plasma was collected from each participant into a 10-ml tube containing 3.2 g/dl sodium citrate. The specimens were centrifuged at room temperature for 15 min at 1,800g. The plasma supernatant was removed with a polypropylene pipette and aliquoted into 0.5-ml plasma samples in polypropylene and stored at −20°C at study site for less than 1 month, then shipped to MUSC and frozen at −80°C until assay. Only plasma samples that were thawed once were used in the analysis. Factor XII:coagulant and HK:coagulant were measured using deficient plasma in a single-stage APTT-based assay using APTT Reagent (Organon Technika, Durham, NC), as previously reported (23). The HK procoagulant assay was performed with total kininogen-deficient plasma (a generous gift of the late Mayme Williams, Philadelphia, PA). Plasma PK was activated with ∼0.4 nmol/l Hageman factor fragment (betaFXIIa), and the formed plasma kallikrein was detected by hydrolysis of the chromogenic substrate H-D-Pro-Phe-Arg-paranitroanilide (S2302) (DiaPharma, Franklin, OH) according to the published procedure, and was expressed as units per milliliter (24). The intra-assay coefficient of variance was 0.85% (24). The plasma PK level in normal subjects determined by using this assay was 1.02 ± 0.18 units/ml (n = 22) (24). Therefore, each unit of PK was defined as the amount of activity associated with 1 ml of pooled normal plasma. Pooled normal human plasma and factor XII−deficient plasma were purchased from George King (Overland Park, KS).

KKS components were measured sequentially in 641 subjects out of the total cohort as they appeared at study sites for scheduled visits. The clinical characteristics of the subjects in which KKS components were measured were compared to those of the remaining EDIC subjects. No differences in age, sex, HbA_1c, BP, AER, DCCT treatment group, duration of diabetes, or BMI were observed between the two groups.

The association among plasma PK, BP (systolic and diastolic), and AER was examined using a multiple linear regression model. BP and AER were treated as dependent variables, whereas plasma PK was used as the main independent variable. Linear regression diagnostics suggested performing a logarithmic transformation of AER to accommodate the nonnormal distribution of AER. Other covariates (age, HbA_1c, duration of diabetes, sex, and DCCT treatment group) were accounted for in all multiple regression models. The square of the multiple partial correlation coefficient was calculated to estimate the increase in the variance of the dependent variable obtained by introducing that variable into a model already including all the others.

Multiple logistic regression models were estimated to assess the association between plasma PK and hypertension status (systolic/diastolic BP ≥140/90 or taking hypertension medication) or between plasma PK and micro-/macroalbuminuria status (AER ≥40 vs. AER ≤40 mg/24 h). The odds ratio (OR) of hypertension and albuminuria per unit increase in PK levels was calculated along with the 95% confidence interval (CI). Goodness-of-fit was evaluated using the Hosmer-Lemeshow χ² statistic. Multiple logistic regression models were adjusted for age, HbA_1c, duration of diabetes, sex, DCCT treatment group, anti-hypertensive medication, and documented hypertension. All statistical analyses were performed using SAS, and the values are expressed as means ± SE.

The circulating levels of plasma PK, factor XII, and HK measured in 636 subjects are shown in Fig. 1. There were no significant differences in the plasma levels of PK, factor XII, or HK between male and female subjects. PK levels were symmetrically distributed among the patients and ranged from 0.2 to 3.0 units/ml, with a mean value of 1.29 ± 0.01 units/ml. Plasma factor XII levels ranged from 0.01 to 9.2 units/ml, with a mean value of 1.06 ± 0.03 units/ml. Plasma HK levels ranged from 0.1 to 5.0 units/ml, with a mean value of 1.34 ± 0.03 units/ml. In cross-sectional analyses, plasma PK levels positively correlated with HK (r = 0.21, P < 0.0001) and XII (r = 0.08, P < 0.04).

Table 1 shows the relation of plasma PK, HK, and factor XII with biochemical parameters in the patient cohort. Univariate analysis showed BMI (r = 0.1, P < 0.01), HbA_1c (r = 0.16, P < 0.0001), and mean BP (r = 0.19, P < 0.0001) to be significantly and positively correlated with PK, whereas body weight (0.08, P < 0.05), waist-to-hip ratio (r = 0.08, P < 0.04), and HbA_1c (r = 0.09, P < 0.02) were significantly and positively correlated with HK.

The results shown in Fig. 2 demonstrate that plasma PK levels positively correlated with diastolic blood pressure (DBP; r = 0.17, P < 0.0001) and systolic blood pressure (SBP; r = 0.18, P < 0.0001; n = 636). The relation between SBP and DBP and plasma factor XII or HK values was not statistically significant (data not shown). A multivariate linear regression model for BP was also constructed (Table 2). PK levels, age, and male sex were significant independent correlates of both SBP and DBP. The semipartial r² value for PK was 2.93% for SBP and 2.92% for DBP (P < 0.0001).

We next analyzed the data categorically with respect to current BP status. Figure 3 demonstrates that type 1 diabetic patients with blood pressure values ≥140/90 mmHg had increased plasma PK levels compared with type 1 diabetic patients with a BP <140/90. Mean plasma PK levels in patients with BP ≥140/90 mmHg were increased compared with the level in patients with BP <140/90 mmHg (1.53 ± 0.07 vs. 1.27 ± 0.02 units/ml; n = 52 and 492, respectively; P < 0.0001). Because some of the hypertensive patients were on ACE inhibitor therapy (ACEI), we examined whether ACEI influenced the level of plasma PK in this patient cohort. The plasma PK levels in patients treated with ACEI were similar to that in patients not treated with ACEI (1.35 ± 0.04 vs. 1.28 ± 0.02 units/ml; n = 92 and 549, respectively; P < 0.145). Among the 549 patients used for the analysis with no ACEI treatment, ∼91% of patients were normotensive and ∼9% were hypertensive and being treated with BP-lowering medications other than ACEI.

The cumulative distribution of plasma PK levels versus hypertension status (all patients diagnosed with hypertension compared with those who remained normotensive) is shown in Fig. 4. The patients who developed hypertension had a significantly higher level of plasma PK compared with patients who did not develop hypertension (1.39 ± 0.04 vs. 1.26 ± 0.02; P < 0.002).

A multiple logistic regression model determined the strength of the association of plasma PK with hypertension. After controlling for age, duration of diabetes, sex, DCCT intensive group, hypertension medications, and HbA_1c, there was a 72% increase in the risk of being hypertensive for every one unit increase in PK level (OR = 1.72, 95% CI 1.09–2.74; P < 0.021).

Because 30–40% of type 1 diabetic patients develop diabetic nephropathy, we determined if PK levels correlated with the degree of AER. On cross-sectional analyses, there was a significant positive correlation between PK levels and log AER (r = 0.16, P < 0.0001) (Fig. 5). Our findings also indicated that diabetic patients with AER ≥300 mg/24 h had a significantly higher level of plasma PK compared with diabetic patients with AER ≤40 mg/24 h (1.45 ± 0.08 vs. 1.27 ± 0.02; n = 34 and 535, respectively; P < 0.01). Diabetic patients with AER 40–300 mg/24 h had an intermediate plasma PK level of 1.38 ± 0.05 mg/24 h (n = 63) (Fig. 6); however, the mean value was not significantly different from that in the normo- and macroalbuminuric subgroups. To further explore whether PK was associated with microalbuminuria, we performed a univariate regression analysis between PK and AER in patients with AER 40–300 mg/24 h and observed a positive and significant correlation (slope = 4.07 + 0.29; r = 0.27, P < 0.03; n = 63). No correlation was observed between plasma PK and AER in normoalbuminuric patients with AER ≤40 mg/24 h (slope = 2.00 + 0.14; r = 0.08, P < 0.06; n = 535). In a multivariate linear regression, plasma PK was independently correlated with log AER (Table 3). The semipartial r² values for PK, age, SBP, DCCT intensive group, and HbA_1c were 0.69, 0.82, 7.09, 2.33, and 5.35%, respectively.

The cumulative distribution of plasma PK by AER is shown in Fig. 7. The data demonstrated that a greater number of patients with AER ≥300 mg/24 h had elevated PK values above any given PK threshold compared with patients with AER ≤40 mg/24 h.

This study represents the first attempt to define the role of the plasma KKS in type 1 diabetes. Our findings provide the first evidence that augmented PK levels positively and independently associate with hypertension and albuminuria, an association that could contribute to the development of vascular disease in type 1 diabetic patients.

The plasma KKS is comprised of a group of plasma proteins that mediate their effects on the vasculature through the release of BK (9,15,16). Recently, it has been shown that endothelial cells express a unique PK activator that converts PK to kallikrein and that factor XII is activated secondary to PK (16,17,25). In our study, we observed a positive correlation between plasma PK and HK as well as PK and factor XII. The increase in PK levels in diabetes may result from either a decrease in PK activation or an increase of its synthesis. Such a process could arise in diabetes, where vascular permeability and reduced vasodilatory responses have been recognized to occur in endothelium. Dysfunction of vascular endothelium may, then, provide a link between albuminuria and atherosclerotic cardiovascular disease (26–28).

An alternative explanation for the elevated PK levels seen in these patients is increased synthesis. In this regard, our findings demonstrated that HbA_1c positively and significantly correlated with plasma PK and HK levels. These epidemiological findings establish an association between metabolic control and KKS components, but do not define the cellular mechanisms or factors that modulate plasma PK synthesis. Because PK is produced mostly in the liver and plasma levels of this protein fall dramatically in liver disease, we have initiated studies to directly assess the role of glucose on plasma PK synthesis in liver cells (29). Previous studies have also reported increased PK levels in patients with diabetes, and in one study (30) an association between plasma PK levels and blood glucose was observed (30,31). Another possible explanation for increased levels of plasma PK in albuminuric patients is that the circulating kallikrein inhibitor, C1 inhibitor, could be reduced via urinary excretion. The smaller molecular mass protein C1 inhibitor (107 kDa) is more likely to be eliminated during glomerular disease than PK, which circulates in a complex with HK with a combined molecular mass of 220 kDa. However, this situation is unlikely because factor XII (80 kDa), which is known to become deficient in nephrotic patients, has a plasma level that positively correlates with plasma PK (32). If reduced C1 inhibitor levels from urinary loss account for an apparent increase in plasma PK, then factor XII levels should also be reduced. Such was not the situation in the diabetic patients examined. Our assay measures activated and nonactivated PK activity and represents total PK mass rather than circulating PK activity. This is independent of C1’s ability to alter PK-specific activity. Thus indirect evidence suggests that the increased PK levels are not caused by artifactual loss of an inhibitor to kallikrein or increased specific activity of the formed enzyme. At this point, however, the information present does not completely exclude increased synthesis in PK in these diabetic patients.

The mechanisms by which diabetes and hypertension interact to accelerate renal damage are as yet undefined. Both conditions are associated with renal endothelial denudation, mesangial cell proliferation, basement membrane changes, and impaired endothelium-dependent relaxation of blood vessels (33,34). Few interventions have been shown to slow the progression of renal disease in diabetic patients. These include intensive glycemic control, blood pressure regulation, and treatment with ACE inhibitors (2,3,8). Despite these interventions, diabetic patients still progress with time to develop end-stage renal disease. The factors responsible for these maladaptive signals leading to end-stage renal failure are still undefined. Our findings indicate that plasma PK levels are positively and independently associated with blood pressure elevation and AER in type 1 diabetic patients. Whether the increase in plasma PK levels we observed in diabetic patients is the cause of injury or the result of injury is yet to be determined.

The renoprotective effects conferred by ACEI could be mediated via reduced conversion of angiotensin I to angiotensin II or via decreased degradation of BK, as ACE is the principal enzyme mediating its catabolism. Recent interventional studies have shown that angiotensin receptor blockers also confer renal protection, suggesting that angiotensin II actions are involved in nephropathy pathogenesis (35,36). Even so, a number of studies have suggested that elevated kinins are responsible for at least some of the protective effects of ACEI (37–39).

We have previously published data indicating that BK and related products of the plasma KKS may play an injurious role as well as a protective role. Indeed, under normal physiological conditions, BK is a potent stimulator of nitric oxide release from renal and endothelial cells, which result in vasodilatation because of effects on underlying vascular smooth muscle cells (40,41). However, when the endothelium is denuded or dysfunctional, as is observed in disease states such as hypertension or diabetes, BK can act directly on mesangial and vascular smooth muscle cells (VSMCs) to induce contraction and to activate multiple signaling pathways, in a manner similar to those activated by vasoconstrictors, ultimately resulting in changes in cell proliferation and increased matrix deposition (42–46). Thus BK can protect from vascular disease associated with diabetes when there is an intact endothelium, but could aggravate vascular disease when endothelium is damaged.

In summary, these investigations demonstrated that plasma PK levels are a marker for vascular disease in patients with type 1 diabetes. The mechanism for the elevation of the plasma PK levels is not known, but could arise from either decreased activation or increased synthesis. Additional studies are underway to determine whether increased PK precedes the rise in blood pressure and/or albuminuria in type 1 diabetes.

This work was supported by a program project grant (PO1-HL55782) cofunded by the National Institutes of Health/National Heart, Lung and Blood Institute and the Juvenile Diabetes Foundation International, National Institutes of Health Grants HL-52779 (A.H.S.) and DK-46543 and HL-71255 (A.A.J.).