Proinsulin Level in Diabetes Mellitus Measured by a New Immunochemiluminometric Assay*

Transcription

1 ANNALS O F CLINICAL AND LABORATORY SCIEN CE, Vol. 25, No. 6 Copyright 1995, Institute for Clinical Science, Inc. Proinsulin Level in Diabetes Mellitus Measured by a New Immunochemiluminometric Assay* TA-JEN WU, M.D.ti, CHING-LING U N, M.D.t, ROBERT L. TAYLOR, B.S.", and PAI C. KAO, Ph.D. 1 Mayo Clinic D epartm ent o f Laboratory Medicine and Pathology, Rochester, MN ABSTRACT Proinsulin, the precursor of insulin and C-peptide, is detectable in the circulation and is a potential m arker of (3-cell dysfunction. Currently, circulating proinsulin can be m easured accurately without the interference of insulin or C-peptide by immunometric assays and a few specific radioimmunoassays. An im m unochem ilum inom etric assay was developed in our laboratory by using two im m unopurified polyclonal antibodies of C-peptide and insulin from two goats. The C-peptide antiserum was labeled with acridinium ester. The insulin antiserum was immobilized onto a plastic bead. W ith this assay, proinsulin levels found in noninsulin-treated patients with noninsulin-dependent diabetes m ellitus (63 ± 58 pmol/l; n = 19) were significantly higher (p < 0.05) than levels in insulin-treated patients with noninsulin-dependent diabetes mellitus (30 ± 24 pmol/l; n = 43), and both w ere significantly higher (p < 0.001) than proinsulin levels in patients with insulin-dependent diabetes mellitus. In addition, there was a negative correlation betw een proinsulin levels and duration of years on insulin therapy in patients with noninsulin-dependent diabetes m ellitus (r = ; p < 0.01). Introduction W ith the recent developm ent in specific proinsulin assays w ithout interference of insulin and C-peptide, it became * Address reprint requests to: Pai C. Kao, Ph.D., Mayo Clinic, 200 First Street, SW, Rochester, MN t Visiting Clinicians at the D epartm ent of Laboratory M edicine and Pathology, Mayo Clinic. t C urrent Address: National C heng Kung University Hospital, Tainan, Taiwan. C urrent A ddress: C athay G eneral H ospital, Taipei, Taiwan. 1 D epartm ent of Laboratory M edicine and Pathology, Mayo Clinic and Foundation, Rochester, Minnesota. clear that the previous understanding of hyperinsulinem ia in patients w ith nonin su lin -d e p e n d e n t d iab e tes m ellitu s (NIDDM) may partly be due to elevated plasma proinsulin levels because most conventional radioim m unoassays m easu rin g im m u n o re a c tiv e in su lin also cross-react w ith p ro in su lin. 1,2,3 ß-cell dysfunction plays an essential role in the pathogenesis of N ID D M.2 Hyperproinsulinemia has been reported in patients with newly diagnosed N ID D M, 4,5,6,7 in islet cell antibody-positive, first-degree re la tiv e s o f p a tie n ts w ith in s u lin - dependent diabetes m ellitus (ID D M ), /95/ $01.20 Institute for Clinical Science, Inc.

2 468 WU, LIN, TAYLOR, AND KAO in women with previous gestational diabetes m ellitus, 9 10 and in patients with im paired glucose tolerance. 11 Therefore, proinsulin has been regarded as a marker of (J-cell dysfunction. Various antidiabetic therapies have different effects on proinsulin levels in N ID D M.5,1 2,1 3 Therapy w ith oral sulfonylurea agents was reported to increase proinsulin levels in N ID D M. 13 H ow ever, proinsulin levels were related to risk factors of cardiovascular disease. 14,15 Human proinsulin could reduce hepatic glucose output in in vivo studies. 16 Subcutaneous injection of hum an proinsulin could norm alize the glucose level in N ID D M 1 6 a n d i n s u lin - d e p e n d e n t (ID D M ). 17 This finding indicates that proinsulin may have a role in the treatm ent of diabetes mellitus, even though an initial trial of proinsulin treatm ent show ed increased risk factors for cardiovascular disease. 18 For further study of the physiologic and pathologic roles of proinsulin, the dem and for m easurem ent of circulating proinsulin is increasing. Currently, proinsulin can be m easured s p e c ific a lly by im m u n o ra d io m e tric assay, 19,20 sensitive and specific radioim m unoassay, enzym e im m unoassay, and im m unofluorom etric assay.23 An im m unochem ilum inom etric assay of proinsulin 24 was developed by us. It was dem onstrated that proinsulin was a good marker for insulinoma, another form of (3-cell dysfunction but with suppressed b lo o d g lu c o se. 2 5 In th is stu d y, our m ethod was tested for detecting proinsulin levels in patients with diabetes who w ere tre a te d w ith th e c u rre n t a n tidiabetic therapy and analyzed the factors a ffe c tin g p ro in s u lin le v e ls in these patients. Patients and M ethods P a t i e n t s In this retrospective study, blood samples o f fasting hyperglycem ic patients (plasma glucose > mg/dl) w ere collected and immediately stored at 70 C. Proinsulin was determ ined in 2 to 3 days. After review of their clinical records, 70 patients w ere found to have diabetes m ellitus: 8 of them ( 2 male and 6 female) had IDDM and were receiving insulin therapy; 62 (29 male and 33 female) had N ID D M. O f th e 62 p a tie n ts w ith NIDDM, 18 were treated with sulfonylurea agents in combination with dietary control by restricting caloric intake, 1 was treated with dietary control only, and 43 were treated with insulin therapy in combination w ith dietary control. T he clinical characteristics of these patients are sum m arized in table I. P atients w ith chronic renal failure who had elevated proinsulin levels owing to im paired renal function25 w ere excluded. P ro in su lin lev els o f p a tie n ts w ith ID D M and those w ith N ID D M who w ere receiving in su lin th erap y w ere com pared to determ ine the difference betw een these two types of diabetes. The results were also compared with those of normal controls.25 In N ID D M, proinsulin levels w ere correlated w ith clinical characteristics, including age, body mass index (weight in kg/height2 in m), blood pressure (systolic and diastolic pressures), duration of diabetes m ellitus, daily dose of insulin, duration of insulin therapy, plasm a gluco se le v e ls, g ly c a te d h e m o g lo b in (HbAlc), and other laboratory data. P a tie n ts w ith N ID D M w ho w ere receiving insulin therapy were compared with those who were not receiving insulin therapy. To study the relationship betw een hyperproinsulinem ia and duration of insulin therapy, quartiles of duration of insulin therapy were used. The duration of insulin therapy (mean ± SD, range) for the first, second, third and fourth quartiles were 1.1 ± 0.7 years, 0 to 2; 4.7 ± 1.2 years, 3 to 7; 8. 8 ± 0.9 years, 8 to 11; and 18.0 ± 4.7 years, 12 to more than 12; respectively. T he proinsulin lev-

3 PROINSULIN LEVEL MEASURED BY NEW IMM UNOCHEMILUMINOMETRIC ASSAY 469 TABLE I Clinical Characteristics of the Studied Diabetic Groups IDDM NIDDM NIDDM on Insulin on Insulin not on Insulin Reference (n -8 ) (n = 43) (n = 19) Ranges Age (years) 41.0 ± ± ±11.0 Sex (male/female) 2/6 15/28 13/6 Duration of DM (years) 28.8 ± ± ± 5.3 Duration of insulin (years) 28.5 ± ± Daily dose of insulin (units) 37.0 ± ± Body mass index 23.7 ± 3-0 (7) 29.9 ± ± 7.1 Systolic pressure (mm Hg) ± 21.0(7) ± ± 19.0 Diastolic pressure (mm Hg) 73.0 ± 13.0(7) 79.0 ± ± Plasma glucose (mg/dl) ± ± ± HbA1 c (%) 12.7 ± 2.5 (7) 11.2 ± 2.6(42) 10.0 ± 2.6(15) 4 7 Creatine (mg/dl) 1.0 ± ± ± Triglycerides (mg/dl) ± 60.0(5) ±138.0(37) ±190.0 (14) <195 Cholesterol (mg/dl) ± 34.0(5) ± 67.0 (38) ± 32.0 (14) <240 HDL-cholesterol (mg/dl) 40.0 ± 11.0(2) 39.0 ± 15.0(22) 33.0 ± 9.0(10) The numbers in parentheses indicate the sample size (in some entries, cases were deleted from studied groups. DM = diabetes mellitus HbA1c = glycated hemoglobin HDL = high-density lipoprotein IDDM = insulin-dependent diabetes mellitus NIDDM = noninsulin-dependent diabetes mellitus els w ere co m p a red w ith th e u p p e r normal lim it of proinsulin, 20 pmol/l,25 and the proportions more than and less th a n th is lim it b e tw e e n q u a rtile s were compared. IM M UNOCHEM ILUM INOM ETRIC ASSAY Proinsulin levels were m easured with a polyclonal immunochem ilum inom etric assay.2 4,2 5 In brief, one goat was im m u nized with synthetic hum an C-peptide. Its antiserum was im m unopurified by a hum an C -peptide affinity column. The p u rifie d C -p e p tid e a n tib o d ie s w ere labeled with acridinium ester and used as signal antibody. The second goat was immunized with pork insulin. Its antiserum was im m unopurified by a hum an insulin affinity colum n. T he purified insulin antibody was immobilized onto a plastic b ead and u sed as capture antibody. During the assay, the bead immobilized w ith insulin antibody was incubated with 250 al of patient s plasma for 5 hours at room tem perature and washed to rem ove C -peptide in plasm a specim en. P ro in s u lin a n d in s u lin w e re captured by the insulin antibody im m o bilized onto the bead. Acridinium esterlabeled signal C -peptide antibody was then added and incubated with the bead for an additional 19 hours. The bead was w ashed again to rem ove excess signal antibody. Only the proinsulin, which has both C-peptide and insulin as part of its m olecule, can bind w ith the signal antibody and be m easured. The w ashed bead was counted in a lum inom eter. R eference range of the assay was 3 to 20 pmol/l. T he assay is specific for proinsulin and has no cross-reactivity w ith insulin or C-peptide.

4 470 WU, LIN, TAYLOR, AND KAO S t a t i s t i c s Mean, standard deviation, and m edian w ere used to express the results. Correlation coefficients were calculated with sim ple linear regression analysis. M ultiple regression analyses w ere performed to examine the relationship betw een proinsulin and clinical variables after logarithm ic transform ation of skew ed data. The M ann-w hitney U test was used to compare the groups. The Kruskal-Wallis test was used to analyze the differences betw een quartiles. The chi-square test was used to test the significance of distribution of proinsulin levels that w ere above or below the upper normal limit. A p value <0.05 was considered significant. Results Patients with IDDM who were receiving insulin therapy had lower levels of proinsulin than patients with N IDDM who w ere receiv in g in su lin therapy. Only 1 patient with IDDM, who had had a pancreatic transplant 1 year previously, had a m easurable proinsulin level of 1 0 pm ol/l. In the 7 other patients w ith IDDM, the proinsulin levels w ere all undetectable. In patients with NIDDM, those without insulin therapy had higher proinsulin levels than those with insulin therapy (figure 1 ). In NIDDM, the proinsulin level had no significant correlation w ith age, duration of diabetes, blood pressure, or body mass index of the patient. The proinsulin O ECL 200 P< P < 0.05 ir 3 W Ç o A. IDDM Mean 2 SD 3 Median <0.6 Range < No. 8 : I* t «ÿ V NIDDM on insulin < NIDDM without insulin CA B-02 FIGURE 1. Proinsulin levels in patients with insulin-dependent diabetes mellitus (IDDM), noninsulindependent diabetes mellitus (NIDDM) with (c) insulin therapy, and NIDDM without (s) insulin therapy. The mean, standard deviation, median, and range are listed under each group.

5 PROINSULIN LEVEL MEASURED BY NEW IMM UNOCHEMILUMINOMETRIC ASSAY 471 level in patients with NIDDM receiving insulin therapy was significantly, but negatively, correlated with duration of insulin therapy (p < ) (figure 2 ). T he lo n g e r th e d u ra tio n of in su lin therapy, the low er the proinsulin level. When the 43 insulin-treated patients with NIDDM w ere divided into 4 quartiles based on their years of insulin therapy, in the first quartile ( 2 or less years), 1 0 p atien ts had a p ro in su lin value > 2 0 pmol/l, and 1 patient had a proinsulin value of <20 pmol/l. However, in the fourth quartile ( 1 2 or more years), only 1 p a tie n t had a p ro in s u lin v alue > 2 0 pmol/l, and 10 patients had a value <20 pm ol/l. T he data strongly suggested that the longer the duration o f insulin therapy, the less frequent the hyperproinsulinem ia. W ith a daily dose of insulin, fasting plasma glucose, triglyceride level, and H D L-cholesterol also show ed significant correlation with the proinsulin level (table II). The proinsulin level also had no correlation w ith glycated hem o globin (HbAlc) in NIDDM with or w ithout insulin therapy. However, with use of m ultiple linear regression analysis, duration of insulin therapy becomes the sole factor that affects the proinsulin level in patients with NIDDM who are receiving insulin therapy. Discussion In this study, the lower proinsulin levels w ere dem onstrated in patients w ith Duration of insulin therapy (yr) Quartile n m r s Proinsulin (pmol/l) >20 <20 No. pt 10 1 No. pt 7 2 Total 11 9 No. pt No. pt CA'168e76B4l8 F i g u r e 2. Relationship betw een proinsulin levels and duration of insulin therapy in quartiles. Precisely 69.4 percent (43 of 62) of patients w ith NIDDM received insulin treatm ent. Their proinsulin levels w ere correlated with the duration of their insulin therapy. The oblique regression line is y = x +44 (p < 0.001, r = 0.5). The vertical dotted line separates the four quartiles of duration of insulin therapy in years. The horizontal line is the upper normal limit of proinsulin, 20 pmol/l. The num ber of the patients w ith a proinsulin level more or less than this limits is listed at the bottom. The difference among quartiles is significant (p < 0.05) by Chi-square test.

6 472 WU, LIN, TAYLOR, AND KAO TABLE II Correlation Coefficients Between Proinsulin and Clinical Characteristics in Noninsulin-dependent Diabetes Mellitus n r p Value Age NS Duration of insulin <0.01 Daily dose of insulin <0.05 Duration of DM history NS Body mass index NS Systolic pressure NS Diastolic pressure NS Plasma glucose <0.05 HbA1c NS Creatinine NS Triglycerides <0.05 Cholesterol NS HDL-cholesterol <0.01 DM = diabetes mellitus NS = no significance HbA1 c = glycated hemoglobin n = number of patients HDL - high-density lipoprotein r = correlation coefficient IDDM who w ere taking insulin therapy. All of these patients have been on insulin therapy for more than 10 years. Only 1 p a tie n t w ith ID D M, who had had a pancreatic transplant 1 year previously, had a proinsulin level of 10 pmol/l. Of the 7 patients w ith IDDM who had an undetectable proinsulin level, 3 suffered from episodes of diabetic ketoacidosis, and 1 had brittle diabetes. Because proinsulin has a pharmacologic action like insulin and is hepatospecific,2 6,2 7 longer acting, 16 and u n d etectab le proinsulin indicates little (3-cell function, hypoproinsulinem ia presum ably has a close relationship to diabetic ketoacidosis and brittle diabetes, which characterize IDDM. It is speculated that the following factors might cause decreased proinsulin levels in IDDM : (1) immunologic damage of (3-cells; (2) exhausted 3-cells owing to an insult of glucose toxicity; 28 and (3) the suppressive effect of exogenous insulin therapy, w hich was also shown in the patients w ith N ID D M, either w ith or without insulin therapy in this study, and was also reported by others.23,29,30 The assay requires two fairly long incubation steps. The first one of 5 hours is for insulin antibodies im m obilized on the plastic bead to recognize and bind the insulin portion of a proinsulin molecule. The bead is then washed to remove any C-peptide m olecule which is present in high concentration in the serum that neutralizes a substantial amount of acridinium ester (tracer) labeled C-peptide antibody added in the second incubation step. The second incubation step of 19 hours is even longer than the first step. The longer incubation step is derived from e x p e rim e n tal titra tio n. D u rin g developm ent of the assay, 2, 4, 6, and 19 hours of incubation, were tested for the tracer C-peptide antibody to recognize the C -p e p tid e p o rtio n of p ro in su lin bound on the bead by immobilized insulin antibody. It was found that 19 hours in the second incubation step gave us the highest sensitivity and best duplication. Thus the long incubation of 19 hours was chosen. Among all the immunoassays, im m unochem ilum inom etric assays are m ost se n sitiv e and en v iro n m en ta lly

7 PROINSULIN LEVEL MEASURED BY NEW IMMUNOCHEMILUMINOMETRIC ASSAY safe.2 4 T h ese conditions can only be m atched by immunofluorometric assays using europium tracers. As sh o w n in o th e r r e c e n t s tu d ie s, 2,4,1 6,1 2 h y p e rp ro in su lin e m ia was dem onstrated in NIDDM, and proinsulin levels were higher in patients who were n o t re c e iv in g in su lin th e ra p y th an in those w ho w ere receiving insulin therapy.2 9 The hyperproinsulinem ia was caused either by 3 -cell dysfunction5,7,8 or by th e pharm acologic effect of sulfonylu rea a g e n ts. 1 3,2 9 Som e stu d ies have reported that fasting proinsulin levels are not correlated with the effectiveness of glycemic control6,3 0 in NIDDM. In our study, fasting proinsulin levels w ere m arginally inversely correlated with fasting plasm a glucose levels, even though they w ere not correlated with glycated hem o globin. H ow ever, there was no correlation betw een plasma glucose and proinsu lin le v e ls in p a tie n ts w ho w e re receiving insulin therapy. A normal proinsulin level, hypoproinsulinem ia, and even an undetectable proin su lin lev e l in som e p a tie n ts w ith NIDDM, especially those with long-term insulin therapy, were found in this study. Similarly, recent studies2 3,2 9,3 0 showed that exogenous insulin m ight suppress hyperproinsulinem ia in NIDDM. Lindstrom et al reported that it might be suppressed to a normal lim it.21 In addition to the relation betw een proinsulin levels and exogenous insulin, it was found by us that the duration of insulin therapy was a more dom inant factor than a daily insulin dose in affecting proinsulin levels in the insulin-treated patients with NIDDM. It is shown in figure 2 that the longer the insulin therapy, the lower the proinsulin levels. In our study, patients who had received insulin therapy for more than 1 2 years had almost normal levels or even undetectable levels of proinsulin. It has been dem onstrated by us that patients w ith N IDDM did not always have h y perproinsulinem ia, particularly with long-term insulin therapy that was closely related to the decreased proinsulin level in patients with NIDDM. Actually, no causal effect could b e d e te r m in e d fro m c o r r e la tio n b e tw e e n variables. In addition, proinsulin has insulin-like pharm acologic action, 1 6,1 7 and th u s th e p o s s ib ility c a n n o t b e excluded that undetectable proinsulin levels in some patients w ith N ID D M m ight be the primary factor of hyperglycemia which leads to the requirem ent of insulin therapy. In conclusion, im m unochem ilum inometric assay of proinsulin is as useful as an im m unoradiom etric assay for d etecting proinsulin levels in diabetic patients. Both hyperproinsulinem ia and hypoproin su lin e m ia w ere found in p a tie n ts treated for diabetes. Patients with IDDM who w ere receiving insulin generally had u n d e te c ta b le p ro in su lin lev e ls. Patients with NIDDM who were receiving insulin therapy had a lower proinsulin lev el than those w ith o u t in su lin therapy. T he p ro in su lin levels w ere m arkedly affected by the duration of insulin therapy rather than by the effectiveness of diabetic control and duration of disease. References 1. Davies MJ, M etcalf J, Gray IP, Day JL, Hales CN. Insulin deficiency rather than hyperinsulinem ia in newly diagnosed type 2 diabetes mellitus. D iabetic Med 1993;10: Tem ple RC, Carrington CA, Luzio SD, Owens DR, Schneider AE, Sobey WJ, H ales NC. Insulin deficiency in non-insulin-dependent diabetes. Lancet 1989;1: Tem ple RC, Clark PMS, Nagi DK, Schneider AE, Yudkin JS, H ales CN. Radioim m unoassay may overestim ate insulin in non-insulindependent diabetics. C lin Endocrinol 1990;32: Make ME, Starr JI, Rubenstein AH. Circulating proinsulin in patients w ith maturity onset diabetes. Am J Med 1977;63: Yoshioka N, Kuzuya T, M atsuda A, Iwamoto Y. Effects of dietary treatm ent on serum insulin and proinsulin response in newly diagnosed NIDDM. D iabetes 1989;38: Schmidli RS, Hagan C, Scott RS, Livesey J, Forbes LV. Plasma proinsulin in recently diag

8 474 WU, LIN, TAYLOR, AND KAO nosed type 2 diabetes mellitus. D iabetes Res Clin Pract 1993;20: Davies MJ, M etcalfe J, Day JL, G renfell A, Hales CN, Gray IP. Im proved beta cell function, w ith reduction in secretion of intact and 32/33 split proinsulin, after dietary intervention in subjects w ith type 2 diabetes m ellitus. Diabetic M ed 1994;11: Spinas GA, Snorgaard O, H ärtling SG, Oberholzer M, Berger W. Elevated proinsulin levels related to islet cell antibodies in first-degree relatives of ID D M patients. D iabetes Care 1992;15: Persson B, Hanson U, Hartling SG, Binder C. Follow -up of w om en w ith previous GDM. Insulin, C -peptide, and proinsulin responses to oral glucose load. D iabetes 1991;40: Dornhorst A, Davies M, Anyaoku V, Hampton SM, E lkeles RS, Beard RW, Johnston DG. Abnormalities in fasting circulating proinsulin concentration in m ild gestational diabetes. Clin Endocrinol 1991;34: Davies MJ, Rayman G, Gray IP, Day JL, Hales CN. Insulin deficiency and increased plasma concentration of intact and 32/33 split proinsulin in subjects w ith im paired glucose tolerance. Diabetic M ed 1993;10: D eacon C F, Schleser-M ohr S, Ballmann M, Willms B, Conlon JM, Creutzfeldt W. Preferential release of proinsulin relative to insulin in non-insulin-dependent diabetes m ellitus. Acta Endocrinologica 1988;119: Davies MJ, M etcalfe J, Day JL, G renfell A, H ales CN, Gray IP. Effect o f sulfonylurea therapy on plasma insulin, intact and 32/33 split proinsulin in subject w ith type 2 diabetes m ellitus. D iabetic M ed 1994;11: Nagi DK, H endra TJ, Ryle AJ, Cooper TM, Tem ple RC, Clark PMS, Schneider AE, Hales CN, Yudkin JS. The relationships of concentrations of insulin, intact proinsulin and split proinsulin w ith cardiovascular risk factors in type 2 (non-insulin-dependent) diabetic subjects. Diabetologia 1990;33: Haffner SM, MykkÄnen L, Stern MP, Valdez RA, Heisserm an JA, Bowsher RR. Relationship of proinsulin and insulin to cardiovascular risk factors in nondiabetic subjects. D iabetes 1993; 42: G lauber HS, H enry RR, Wallace P, Frank BH, Galloway JA, Cohen RM, Olefskey JM. The effects of biosynthetic hum an proinsulin on carb o h y d rate m etab o lism in n o n -in su lin - dependent diabetes mellitus. New Engl J Med 1987;316: S pradlin C T, G allow ay JA, A nderson JH. Apparent increase in coronary heart disease w ith hum an proinsulin (H PI) R andom ization failu re or H P I effect? D ia b e to lo g ia 1990;33(Suppl):A Davidson JK, Satterfield LD, Rolfes RM. rdna hum an proinsulin produces excellent m etabolic control of type 1 (insulin-dependent) diabetes. Diabetologia 1987;30:511A. 19. Sobey WJ, B eer SF, Carrngton CA, Clark PMS, Frank BH, Gray IP, Luzio SD, O wens DR, Schneider AE, Siddle K, Tem ple RC, Hales CN. Sensitive and specific two-site immunoradiometric assays for hum an insulin, proinsulin, split and split proinsulins. Biochem J 1989;260: Chevenne D, Ruiz J. Lohmann L, L audat A, Leblanc H, Gray P, Passa P, Porquet D. Immunoradiometric assay of hum an intact proinsulin ap p lied to p atien ts w ith ty p e 2 d iabetes, im paired glucose tolerance, and hyperandrogenism. Clin Chem 1994;40: Bowsher RR, Wolny JD, Frank BH. A rapid and sensitive radioimmunoassay for the m easurem ent of proinsulin in hum an serum. D iabetes 1992;41: Kjems LL, Roder ME, D inesen B, H artling SG, Jorgensen PN, B inder C. H ighly sensitive enzyme immunoassay of proinsulin immunoreactivity with use of two monoclonal antibodies. Clin Chem 1993;39:214& Lindstrom TH, Arnqvist HJ, von Schenck HH. Effect of conventional and intensified insulin therapy on free-insulin profiles and glycemic control in NIDDM. D iabetes Care 1992;15: Kao PC, Taylor RL, H eser DW. C -peptide im m unochem ilum inom etric assay developed from two seemingly identical polyclonal antisera. Ann Clin Lab Sci 1992;22: Kao PC, Taylor RL, Service FJ. Proinsulin by im munochem iluminometric assay for the diagnosis of insulinoma. J Clin Endocrinol Metab 1994;78: Revers RR, Henry R, Schm eiser L, Kolterman O, Cohen R, Bergenstal R, Polonsky K, Jaspan J, Rubenstein A, Frank B, Galloway J, Olefesky JM. The effects of biosynthetic hum an proinsulin on carbohydrate m etabolism. D iab etes 1984;33: Davis SN, Monti L, Piatti PM, Brown M, Hetherington C, Orskov H, Sobey W, H ales CN, Alberti KGMM. Dose-response characteristics of human proinsulin and insulin in non-insulind ependent diabetic hum ans. Am J Physiol 1992;263:E Unger RH, G rundy S. H yperglycem ia as an inducer as w ell as a consequence of im paired islet cell function and insulin resistance: Im plications for the managem ent of diabetes. Diabetologi 1985;28: C ohen RM, N athan DM, C lem ents Jr, RS. Hyperproinsulinem ia in type II diabetes. D iabetes Care 1992;15: Jain SK, Nagi DK, Slavin BM, Lumb PJ, Yudkin JL. Insulin therapy in type 2 diabetic subjects suppresses plasm inogen activator inhibitor (PAI-1) activity and proinsulin-like molecules independently of glycaemic control. D iabetic M ed 1992;10:27-32.