1 Diab Cotnp, 6:116-122

Relationship of Lipid Peroxides to Diabetic Complications Comp&ison With Conventional Laboratory Tests Donald Armstrong Nabella Abdella Amani Salman Nicholas Miller Ehad Abdel Rahman Michael Bojancyzk

ABSTRACT

One hundred and seventy-three Arab patients with non-insulin-dependent diabetes mellitus (NIDDM) and 51 controls were examined. The routine laboratory tests of fasting blood sugar (FBS), glycosylated hemoglobin (HbA1,l, fructosamine (FRA), and triglycerides (TG) were compared with lipid hydroperoxides (LHP) as detected by the thiobarbituric acid (TBAl method, which measures total TBA reacting substances (TBARS). All routine laboratory measures were elevated in the patient population. In the case of LHP, values in NIDDM were 491 + 183 PmollL for men and 507 f 183 Ir.mol/L for women, as compared with pooled normal values of 275 f 85 PmollL. Seven patients with NIDDM undergoing hemodialysis were also elevated above normal (441 + 97 pmol/L). Comparisons between LHP and the other biochemical parameters indicated a positive statistical correlation with triglycerides (p < 0.0011; FBS, HbA1,, and FRA (p > 0.011. Patients with NIDDM on dietary control had somewhat lower LHP levels (441 + 135 pmol/Ll than those receiving insulin (519 f 206 t.r,mol/Ll, or oral hypoglycemic agents (518 f 177 pmol/Ll. Dietary maintenance also resulted in reduced FBS and HbA1, levels. Subdivision of patients according to duration of disease and by nationality showed that Kuwaiti nationals had slightly higher LHP values

than did patients of other Arabic origin. Women patients (Kuwaiti and non-Kuwaiti) with NIDDM disease duration of greater than 10 years, had LHP levels above those of males (551 + 245 versus 478 f 126 pmol/L) and, as a group, showed significant enhancement of retinal complications (85.5% versus 55%). A statistically significant trend (p < 0.081 could be demonstrated between the prevalence of retinopathy and increasing serum LHP levels. A similar relationship was also noted for proteinuria, but not for macrovascular complications. When the effect of glucose was examined, we noted that prevalence was the same as the prevalence of patients who had a TBARS value in the ~500 pmol/L range. Furthermore, these rates did not change when patients were markedly hyperglycemic (>lO ~mol/Ll. The results of the studies reported here indicate a correlation between LHP and tissue damage in diabetes, especially in the retina and kidney. Patients with these complications had LHP levels almost two times higher than normal, and thus, it is proposed that, on the basis of these results, the TBARS assay be considered an important adjunct to the battery of laboratory tests currently ordered by the physician. This might prove useful in both diagnosis and management. (The Journal of Diabetes and Its Complications 6:116-122, 1992.1

0891-6632/92/$5.00 0 1992 Journal of Diabetes and Its Complications

] Diab Comp 1992; 6:2 LIPID PEROXIDES AND DIABETIC COMPLICATIONS 117

INTRODUCTION

D uring the past few years, several biochemi- cal and pathological findings have been re- ported to correlate with the clinical features of non-insulin-dependent diabetes melli-

tus (NIDDM). More recently, increased peroxidation of lipids has received a great deal of attention, and studies now provide evidence to suggest a role for lipid hydroperoxides (LHP) in the pathology of dia- betic complications. ‘r2

Yagi3 was the first to demonstrate that LHP were present in normal human serum and transported via low-density lipoproteins. These investigators found elevated levels in the whole serum of patients with NIDDM, and this was associated with a shift to the high-density lipoprotein fractiona In vivo and in vitro experiments have shown that lipoproteins containing LHP bind to the vessel wall and provoke increased lipid deposition within endothelial and smooth mus- cle cells.5 Animals made diabetic with alloxan or strep- tozotocin (STZ), also have increased levels of serum LHP, as well as increased deposition of lipids in vas- cular endothelium.6,7 Induced diabetes likewise pro- vokes alteration of the electroretinogram.s*9

The hyperglycemia associated with diabetes causes nonenzymatic glycosylation of proteins and hemoglo- bin. Two laboratory assays currently employed for the diagnosis and management of diabetes are determina- tion of the major glycosylated hemoglobin (HbA1,) fraction and fructosamine (FRA), which are measures of total serum or plasma glycosylated proteins. Both provide information about glycemic control, but re- flect changes over a period of weeks or months. Be- cause of the time factor, the high cost per test and discrepancies arising from interfering substances or differences in methodology, the clinical utility of fruc- tosamine measurements still remains to be estab- lished and fully accepted by the hospital labora- tory. *“,ll HbAI,, however, remains a well-established test.12

To assay for LHP, the most common method em- ployed is to acid hydrolyze all lipid peroxides to ma- londialdehyde (MDA) and then measure the product after complexing with thiobarbituric acid (TBA). Be- cause a small amount of MDA may arise from other sources, the assay is more correctly referred to as mea-

- Reprint requests to be sent to: Dr. Donald Armstrong, Depart-

ment of Medical Technology, State University of New York (SUNY) at Buffalo, UB Clinical Center, 462 Grider Street, Buffalo, NY 14215.

Department of Medical Laboratory Technology (D.A., N.M., E.A.R.) and Health Information Administration (M.B.), Faculty of Allied Health Sciences and Nursing, Kuwait University, and De- partment of Medicine (N.A., AS.), Kuwait University Health Sci- ences Center Diabetes Clinic, Mubarak Al-Kabeer Teaching Hospi- tal, Ministry of Public Health, Kuwait.

suring TBA reacting substances (TBARS). Studies using this method have shown that LHP were signifi- cantly higher in patients exhibiting angiopathic changes, i.e., retinopathy and atherosclerosis, as compared with diabetes patients without clinical manifestations.3,12 LHP are formed in membranes by enzymatic and nonenzymatic pathways and partici- pate in the propagation of sustained-membrane dam- age via free radical intermediates.13 Another source of free radicals is the interaction of LHP with glucose and glycosylated peptides. l4

Preliminary studies by our laboratory reported that serum LHP as measured by TBARS, were elevated in a small group of 20 Arab patients with NIDDM.15 The objective of the present study was to evaluate further the usefulness of the TBARS assay in a larger group of patients with NIDDM, and evaluate the correlation between TBARS level with the prevalence of retinopa- thy, proteinuria, and macrovascular complications.

SUBJECTS AND METHODS

Study Group. A total of 86 men and 80 women of Arabic origin with an established diagnosis of NIDDM according to the American Diabetes Association Clini- cal Practice Recommendations”j were selected from the Diabetes Clinic at the Mubarak Al-Kabeer, Univer- sity of Kuwait Teaching Hospital over an ll-month period from June 1989 to May 1990. The average num- ber of visits was 6 times per year. Of the 186 patient total, 76 (41%) were Kuwaiti nationals (34 men and 43 women); the remaining 90 patients (52 men and 38 women) were a mixture from Jordan (19%), Egypt (9%), Iraq and Syria (5% each), Lebanon and the Sudan (4% each), and Oman, Palestine, and Yeman (3% each). The majority of patients were obese with a body mass index of greater than 28 for men and greater than 32 for women. Normal values for this calculation are in the range of 20-25 for men and 19-24 for women.17 The duration of the disease ranged from 1 month to 27 years, the average being 8.7 years. Dietary and treatment compliance was poor and only 35%-40% of the patients used self-monitor- ing. Control serum was obtained from Arabic men and women employees from the hospital chemistry laboratory to establish mean values and calculate the 2 + standard deviation. These 51 subjects were deter- mined as normal on the basis of an unremarkable SMAC-20 chemistry profile and no history of disease.

Diagnostic Features of Complications. Proteinuria of greater than 250 mg/24 h was the criteria employed for establishing a diagnosis of renal dysfunction and a history of angina pectoris, abnormal exercise findings, and evidence of myocardial infarcts supported by electrocardiogram findings was the criteria for cardio- vascular disease. Elevated blood pressure, a history

118 ARMSTRONG ET AL.

of documented transient cerebral ischemia, or estab- lished cerebrovascular accident was the basis for a di- agnosis of macrovascular disease. Seven additional patients with end-stage renal disease (ESRD) on a thrice weekly maintenance hemodialysis program were also studied. A diagnosis of retinopathy was made on the basis of retinal hemorrhage present in one or two quadrants, cotton wool spots, venous cali- ber change including heading, intraretinal microvas- cular abnormalities, numerous microaneurisms and exudates, or evidence of proliferative disease with fi- brovascular traction resulting from neovasculari- zation.

Methods. Whole blood was collected into vacu- tainer tubes and centrifuged to obtain serum for the determination of fasting blood sugar (FBS) and triglyc- erides (TG), which were determined on the SMAC- 20. The range for normal in the routine hospital was used. The percentage of HbA1, in red cell hemolysates was determined using a commercial kit (Glytrac) mar- keted by Corning Medical Division. Samples were subjected to electrophoresis, stained, and the amount of HbAi, relative to unglycosylated hemoglobin was quantified by densitometry. The mean, SD, and CV on quality control sera from ASCP Check Samples were similar to those reported by the manufacturers. Fructosamines were measured calorimetrically by the reduction of nitroblue tetrazolium (NBT) to a blue di- formazan derivative using the Hoffman-LaRoche kit. A Union Carbide Centrifichem 500 autoanalyzer was employed to quantify the reaction, with pipettor set- tings at 30 PL for sample, 70 FL for diluent, and 350 PL for reagent. After 5 min of incubation at 37”C, the reaction was monitored by the change in absorbance at 550 nm. The correlation coefficient of this test against an high-performance liquid chromatography (HPLC) reference method was 0.98.17 Serum TBARS was determined using a modified fluorimetric assay where the MDA-TBA adduct is formed by acid hy- drolysis at 100°C and quantified by setting the excita- tion wavelength at 510 nm and measuring the emitted light at 553 nm as previously described.i5 Measure- ments were taken on a Shimadzu RF5000 recording

J Diab Comp 1992; 6:2

spectrofluorophotometer, and the data plotted with a log x-axis and logit y-axis where the curve is described by a third degree polynomial. The standard used for the TBARS assay was oxidized 1,1,3,3-tetraethoxypro- pane, and the assay was linear over the range of 0.3-1.25 pmol/L.

Colored photographs were taken of patients in the Diabetes Clinic with a computerized Nikon Fundus Camera system.

The collected data were entered and stored in dBASE IV format to allow for easy storage, manipula- tion, and retrieval. All the statistical analyses have been performed using the comprehensive and indus- try standard software package, SPSS/PC + version 4.0 (SPSS Inc., Chicago, IL, 1990). Preliminary descriptive statistics were run in order to calculate frequencies of occurrence for categorical data, and mean and stan- dard deviations for the ordinal data. The level of at least p < 0.01 (2-sided) was taken as significant. The t test procedure was run to analyze the statistical sig- nificance of the difference of mean values of interest for the chosen samples. In most cases, the exact value of the probability of rejection of the null hypothesis under standard assumptions has been calculated. This p value has been given in the tables only when the value is low enough to be able to reject the hypoth- esis of no difference. A linear-regression analysis was used to determine possible relationships between TBARS level and type of complication.

RESULTS

The mean age of patients with NIDDM was 52 ? 10 years for men and 51 ? 10 years for women. There were 85 patients with disease duration of less than 10 years (mean, 4.3 years) and 81 patients with disease duration of greater than 10 years (mean, 15.1 years). Patients were generally hypertensive (mean, 135 + 19 mm Hg systolic and 82 + 8 mm Hg diastolic) and not well controlled as evidenced by the high FBS, HbA1,, and FRA values (Table 11. TG values were also above normal. A marked increase was observed in the level of TBARS for men (491 + 183 pmol/L) and women (507 f 183 pmol/L), which was significant at

TABLE 1. BIOCHEMICAL MEASUREMENTS IN NON-INSULIN-DEPENDENT DIABETES MELLITUS (NIDDM)

Age FBS Hb AI, TG TBARS Patient Group n (years) (mmoYL) (%) CmmFRoAvu (mmoI/L) (pmoI/L)

Control 51 35 c 15 5.1 2 0.3 4.6 2 0.3 0.42 it_ 0.07 1.6 ” 0.2 274.9 k 84.7

NIDDM Male 86 52.3 2 9.5 9.1 If: 3.0 10.5 k 2.5 0.63 f 0.11 2.3 + 1.8 490.6 f 183.2* Female 80 50.9 f 10.2 10.1 +- 4.1 10.5 t 2.5 0.58 5 0.11 2.4 f 1.8 506.6 + 182.7

* p < 0.0001. FBS, fasting blood sugar; Hb&, glycosyhted hemoglobin; FRA, fructosamine; TG, triglycerides; TBARS, thiobarbituric acid reacting substances.

] Diab Comp 1992; 6:2 LIPID PEROXIDES AND DIABETIC COMPLICATIONS 119

TABLE 2. COMPARISON OF BIOCHEMICAL MEASUREMENTS AS A FUNCTION OF TREATMENT IN NON-INSULIN-DEPENDENT DIABETES MELLITUS

Patient Group Age FBS TG TBARS

n (years) (mmoYL) (mmoI/L) HbAl, (pmol/L)

Diet Oral hypoglycemic

agents Insulin

24 51.5 5 10.6 6.8 k 1.5 2.1 + 1.8 8.6 k 2.0 0.57 t 0.09 440.9 + 134.8 86 53.0 k 10.8 9.3 ” 3.1* 2.2 + 1.4 10.4 k 2.3* 0.60 ? 0.10 518.5 k 177.3

51 51.2 i 9.2 11.3 2 4.2* 2.5 k 2.2 11.9 2 2.3* 0.68 + 0.12 494.7 r 195.3

* p < 0.002. FBS, fasting blood sugar; TG, triglycerides; HbAI,, glycosylated hemoglobin; FRA, fructosamine; TBARS, thiobarbituric acid reacting substances.

p < 0.0001. The predialysis mean value of TBARS for patients with a combination of NIDDM and ESRD, was 441 rt 97 p,mol/L. Correlations comparing the var- ious biochemical parameters chosen in this study for NIDDM, were positive between TBARS, FBS, TG, and HbAr,; between FBS, HbAr,, and FRA; and between HbAr, and FRA. The strongest significance (p <

0.001) was found for TBARS and TG, with FBS and HbA1, showing a lower significance (p < 0.01). In contrast, there were no correlations noted between FBS and TG.

Because patients with NIDDM in this study were being managed on different treatment regimens, we examined whether there was any effect upon their standard biochemical profile and LHP level (Table 2). A total of 53% of the patients in this study were taking an oral hypoglycemic agent (OHA), another 32% re- quired insulin therapy, and 15% of the patients were controlled by dietary means. There were no statisti- cally significant age differences in the patients among the groups. The FBS level was only slightly above nor-

mal in the diet group, whereas in the OHA group it was similar to the overall value shown in Table 1 and was highest (11.3 + 4.2 wmol/L) in the insulin-treated group. TG values were elevated in all three groups. TBARS were lowest in the diet group (441 + 135 kmol/L) and highest in the OHA group (519 + 177 pmol/L). The insulin-treated group was intermediate at 495 + 195 PmoYL. The Fprob for TBARS between these three groups was 0.2974 and therefore not sig- nificant.

Figure 1 shows the relationship between LHP level and prevalence of each major complication. In this illustration, levels of TBARS were set arbitrarily with the lowest cutoff value (~500 pmol/L) corresponding to the mean of all patients as shown in Table 1 and then set at 100 PmoYL increments. Of 103 patients with values below 500 tJ,mol/L, 28 (27%) had retino- pathic changes that increased in relationship to TBARS, i.e.; 15 (37%) of 41 patients had retinopathy coinciding with values between 500 and 600 pmol/L and 9 (50%) of 18 patients had retinopathy coincident

PREVALENCE(%) 804, I

<500 500-600 600-700 700-600 >600

TBARS (pmol/L)

FIGURE 1 complication .

Relative fhioburbifuric acid reacting substances (TBARS) acfiuify us a function of

120 ARMSTRONG ET AL. j Diab Comp 1992; 6:2

TABLE 3. COMPARISON OF BIOCHEMICAL MEASUREMENTS AND CLINICAL COMPLICATIONS FOR KUWAIT1 NATIONALS AND OTHER ARAB PATIENTS WITH NON-INSULIN-DEPENDENT DIABETES MELLITUS

Macrovascular ABe FBS TG HbAlc FRA TBARS Retinopathyt Proteinuriat Disesset

Patient Group n (yea& (mmcdiL) (mmoUL) (%6) fmmol/L) (pmol/L) BhW f%) (%I (9s)

Males Diabetes duration

Cl0 years

Kuwaiti (KU) 23 51.0 r 10.6 9.3 + 3.1 2.5 f 2.1 10.0 + 2.6 0.63 e 0.11 585 2 250 28.6 2 3.7 22.7 14.3 30.4 Non-Kuwaiti (NK) 31 52.7 + 11.8 8.5 ? 2.7 2.0 + 1.0 10.2 + 2.2 0.63 e 0.07 443 2 143’ 29.3 + 3.7 12.5 6.3 40.6

Diabetes duration

>lO year

Kuwaiti 11 55.2 + 5.7 9.5 f 4.1 2.5 ‘- 2.2 10.9 k 2.2 0.58 + 0.13 510 -c 108 28.7 2 5.7 54.5 18.2 63.6 Non-Kuwaiti 20 50.5 + 10.5 9.7 ‘- 3.0 2.6 -c 2.1 11.5 c 3.3 0.64 + 0.16 447 -c 145 27.1 5 4.2 55.0 20.0 57.9

Females Diabetes duration

<IO years

Kuwaiti 27 49.3 f 9.4 9.4 t 4.1 2.4 k 1.3 8.7 + 1.7 0.52 f 0.07 490 ? 112 32.7 k 5.7 12.5 11.5 34.6 Non-Kuwaiti 19 50.6 + 9.6 9.9 + 2.5 2.6 2 2.8 10.4 + 2.5b 0.57 + 0.12 451 + 105 33.5 -t 5.3 22.2 5.6 11.1

Diabetes duration

>I0 year

Kuwaiti 15 53.8 of: 11.2 12.0 -c 5.1” 2.3 -c 1.4 11.6 e 2.J 0.60 2 0.10 553 + 195 30.0 ? 6.0 77.8 10.0 40.0 Non-Kuwaiti 20 56.5 -c 9.3 10.1 2 4.8 2.1 k 1.7 11.2 k 2.7 0.66 k 0.10 549 + 283“ 30.8 + 5.7 93.3 26.7 56.3

FBS. fasting bled glucosp; TG. triglycerides; HbAk. glycosylated hemoglobin; FRA, fructusamine; TEARS, thiobarbituric acid reacting substances.

l BMI = weight (kg) + J@$%?

t W of patients with clinical findings.

’ The diference between FBS values for KU females as R function of duration is p < 0.007.

b The difference between HbAk values for KU and NK femnles with duration of <IO yr, IS p < 0.005.

‘ The diffmnce ktrveen HbAk unlues for KU and NK males with duration of ~10 yr. is p 4 0.007

d The difference between TBARS vnlues in NK females RS R function of duration is p i 0.099.

with values between 600 and 700 kmol/L. This trend continued into the higher TBARS range, with 5 (56%) of 9 patients having retinopathy present at concentra- tions of 700-800 pmol/L and 7 (70%) of 10 patients having serum levels exceeding 800 pmoI/L. The four highest levels recorded were 1568,1526,1123, and 908 pmol/L. The relationship between TBARS and pro- teinuria was most noticeable only above the 600-700 kmol/L level. The prevalance of macrovascular dis- ease was higher than retinopathy at the <500 pmol/L level where 38.5% of the patients exhibited this com- plication, but further increases did not occur until the MDA values exceeded 700 PmollL. This data was fur- ther analyzed to determine the significance of the trends illustrated in Figure 1. With TBARS as the inde- pendent variable and type of complication the depen- dent variable, the level of statistical significance for the regression coefficients was p < 0.080 for retinopa- thy, p < 0.001 for proteinuria, and p < 0.133 for macro- vascular disease, which is not significant.

Because complications associated with NIDDM are more prevalent in patients having the duration of the disease over 10 years, we evaluated the incidence of retinopathy, renal disease, and macrovascular disease as a function of time (Table 3). We also factored out sex and ethnic differences between Kuwaiti and other Arabic patients and found that Kuwaiti men had higher levels of LHP and TG than their counterparts and changed little with duration. Women, on the

other hand, had higher levels of FBS, HbArc, and TBARS than men when disease had been present for more than 10 years. Otherwise, no major shift was noted between the Kuwaiti and non-Kuwaiti women. It can be readily seen from this table that an increased prevalence of retinopathy occurs after 10 years of dis- ease duration in both cohorts, and that women are the group most affected. In men, macrovascular dis- ease was roughly equal to retinopathy in the over lo- year group. Both groups were overweight, with women showing the greater calculated body mass index (BMI) and upper fat distribution. Obesity was not much different between the under lo-year group and the over lo-year group.

DISCUSSION AND CONCLUSION

The results obtained for LHP as measured by the TBARS test in a large population of Arabic patients are in agreement with other nationalities.1,3*6 In our group of poorly controlled NIDDM patients whose glucose and I-&Al, levels were markedly elevated, the amount of circulating TBARS was essentially twice the control value. When patients were segregated accord- ing to sex and duration of disease, we found a marked increase in the level of TBARS in women with NIDDM duration of greater than 10 years. Further analysis of the data revealed that Arabic women had a singularly overall higher incidence of retinopathy (87.5%) than did the men (54.8%). When the women were subdi-

1 Diab Comp 1992; 6:2

vided according to nationality, the prevalence was found to be an astounding 93.3% in non-Kuwaiti Arab patients; however, the Kuwaiti women were also af- fected to a high degree (77.8%), which again, was greater than observed in men (54.5%). Prior investiga- tions at Kuwait University Health Sciences Center have reported on the elevation of lipids in women with NIDDM.i* Others have found that reduced es- trogen levels, as may be expected in women after menopause, contribute to increased levels of LHP.19 The age of patients in our study would be consistent with that finding.

An initial report by our group suggested a relation- ship between high TBARS values and retinopathy.15 That study examined 20 patients of mixed sex and showed a strong relationship between TBARS and presence of retinopathy, however, the sample size was small. We have now been able to confirm this observation in a substantial number of patients and have documented that when the serum level is 500 mmol/L (the mean value) or less, the prevalence of retinopathy is 27%. Moreover, as the serum level in- creases, so does the rate (Figure 1). For example, when TBARS was between 500 and 600 PmollL, the prevalence of retinopathy rose to 37%, and when the concentration exceeded 600 pmol/L, the prevalence increased accordingly. Consequently, when TBARS values were over 800 kmol/L, the prevalence rate was 70%. This is higher than what was observed when TBARS were compared to the presence of proteinuria or macrovascular disease and emphasizes the sensi- tivity of the retina to peroxidation. Another relation- ship was noted in this regard when complication rate was compared among 34 patients having the disease for more than 10 years. In this evaluation on a group whose duration ranged from 14 to 27 years, we found that 79% had retinopathy, 18% had proteinuria, and 56% had macrovascular findings. By comparison, those with disease duration of less than 10 years showed a strikingly lower prevalence for retinopathy of just 16%. Proteinuria (8.5%) and macrovascular dis- ease (29%) were also lower in this group. Thus, five times more patients had retinopathic change once their disease was extended beyond 10 years, whereas the rate of renal and vascular disease only increased by a factor of two.

The importance of the TBARS findings is shown by examination of prevalence rates as a function of glucose control. When glucose levels were within the normal range, ” the number of patients exhibiting reti- nopathy, proteinuria, or macrovascular disease was about the same as those observed in the TBARS group when values were ~500 pmol/L, i.e., 30%, 9% and 39%, respectively (see Figure 1, for comparison). When glucose levels were out of control (>lO nmoli L) the rates were 41.5%, 20%, and 42%.

LIPID PEROXIDES AND DIABETIC COMPLICATIONS 121

As expected, statistical correlations of laboratory analyses were significant between FBS and HbAi, and between HbA1, and FRA. The strongest correlation, however, was found between LHP and TG (p < 0.001). A similar LHP to TG correlation has been re- ported in elderly diabetes mellitus patients.” In that study, lipid peroxidation was significantly increased in patients not only with clinical evidence of retinopa- thy and renal impairment as we have noted, but with macrovascular disease as well. Thus, our recommen- dation to the diabetologist would be to include the TBARS test for LHP and TG in the diagnosis and man- agement of patients with NIDDM. The TBARS test is inexpensive, sensitive, and easily automated.

A recent review article22 has put forth arguments for the relevance of lipid-protein oxidative stress as an independent risk factor associated with vascular complications in diabetes. In this study, we have doc- umented that aberrations in lipid peroxide concentra- tion appear to be strongly related to the presence of clinical complications which affect the retina and the kidney, however, a correlation with vascular disease did not attain statistical significance. The reason for this is unclear, but may be due to the heterogeneity found in a diagnostic category encompassing macro- vascular changes.

ACKNOWLEDGMENT

This work was conducted at the Faculty of Allied Health Sciences and Nursing, Health Sciences Center, Kuwait Uni- versity, and was supported by grants MDH 183, MH 010, and a general facility equipment grant, MLH 036. Expert technical assistance was provided by Mrs. Reem Younis, Miss Ruba Mohaisen, and Mr. Ahmad Al-Jada, We thank Bertel Berg, M.D., Ph.D., and Richard Lanham, M.D., for valuable consultation, and Maurizio Trevasan, M.D., Ph.D., for the statistical analysis of data in Figure 1.

1.

2.

3.

4.

5.

6.

REFERENCES

Wolff S: The potential role of oxidation and its compli- cations: novel implications for theory and therapy, in Crabbe MC (ed), The Diabetic Complications: Scientific and Clinical Aspects. London, Churchhill-Livingstone, 1987, pp. 167-220.

Oberley L: Free radicals and diabetes. Free Radic Biol Med 5:113-124, 1988.

Yagi K: Assay for serum lipid peroxide level and its clinical significance, in Yagi K (ed), Lipid Peroxides in Biology and Medicine. New York, Academic Press, 1982, pp. 231-232.

Nishigaki I, et al: Lipid peroxide levels of serum lipo- protein fractions of diabetic patients. Biochem Med Metab Biol 25:373-378, 1981.

Yagi K, et al: Formation of lipid-laden cells from cul- tured aortic smooth muscle cells and macrophages by linoleic acid hydroperoxide and low density lipopro- tein. I Girl Biochem Nuts 3:87-94, 1987.

Nakimura H, Kaiho S, Mizuno N: Hyperlipoperoxide-

122 ARMSTRONG ET AL. J Diab Comp 1992; 62

10.

11.

12.

13.

14.

mia in mice induced by alloxan. Rinsho Byori 29:256-261, 1981.

Murata T, et al: Lipid peroxidation in diabetic rat ret- ina. Metab Ped Ophthalmol 5~83-87, 1981.

Pautler E, Ennis S: The effects of induced diabetes on the electroretinogram components of the pigmented rat. Invest Ophthalmol Vis Sci 19:702-705, 1981.

Mac Gregor L, Matschinsky F: Treatment with aldose reductase inhibitor or with myoinositol arrests deterio- ration of the electroretinogram in diabetic rats. ] Clin invest 76:887-889, 1985.

Ashby J, Frier B: Is serum frustosamine a clinically use- ful test? Diabetic Med 5:118-121, 1981.

Buysschaert M, et al: Serum fructosamine level as a marker of glycemic control in diabetic patients with and without a residual C-peptide secretion. Diabetes Res 6:129-131, 1987.

Jennings P, et al: Increased diene conjugate in diabetic subjects with microangiopathy. Diabetic Med 452-456, 1987.

Gutteridge J, Westermarck T, Halliwell B: Oxygen rad- ical damage in biological systems. In Johnson J, Wal- ford R, Harman D, Miquel J (eds), Free Radicals, Aging and Degenerative Diseases. New York, Alan R. Liss, 1985, pp. 99-139.

Hicks M, Delbridge L, Yue D, Reene T: Catalysis of lipid peroxidation by glucose and glycosylated colla- gen. Biochem Biophys Res Commun 151649-655, 1988.

15.

16.

17.

18.

19.

20.

21.

22.

Abdella N, Al-Awadi F, Salman A, Armstrong D: The TBA assay as a measure of lipid peroxidation in Arab patients with NIDDM. Diabetes Res 15:173-177, 1991..

Thomas A, McKay D, Cutlip M: A monograph method for assessing body weight. Am I Clin Nutr 29:302-304, 1976.

Schleicher E, Wieland 0: Specific quantification by HPLC of protein (lysine) bound glucose in human serum albumin and glycosylated proteins. ] Clin Chem C/in Biochem 19:81-87, 1981.

Al-Muhtaseb N, Al-Yusuf A, Abdella N, Feneck F: Lip- oproteins and apolipoproteins in young non-obese Arab women with NIDDM treated with insulin. Diabe- tes Care 12:325-331, 1989.

Nakagawa Y, et al: Effects of ovariectomy on serum and liver lipid peroxide levels in female mice. 1 Clin Biochem Nutr 6:87-94, 1989.

American Diabetes Association: Office guide to diag- nosis and classification of diabetes mellitus and other categories of glucose intolerance: Clinical practice rec- ommendations, American Diabetes Association. Dia- betes Care 14(Suppl 2):3-4, 1991.

Mooradian A: Increased serum conjugated dienes in elderly diabetic patients. 1 Am Geriatr Sot 39:571-574, 1991.

Baynes J: Role of oxidative stress in development of complications in diabetes. Diabetes 40:405-412, 1991.