J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from J Clin Pathol 1984;37:659-664

Platelet and factors in proliferative diabetic retinopathy

DQ BORSEY,* CV PROWSE,t RS GRAY,* J DAWES,t K JAMES,§ RA ELTON, | BF CLARKE* From the *Diabetic and Dietetic Department; tBlood Transfusion Service, Royal Infirmary, Edinburgh; tMRC Immunoassay Team, Edinburgh; §Department ofSurgery and || Medical Computing and Statistics Unit, University ofEdinburgh, Scotland

SUMMARY Plasma ,8-thromboglobulin, factor 4, , fibrinopeptide A, antithrom- bin III, factor VIII related antigen, a2-macroglobulin, platelet count, and total glycosylated haemoglobin were measured in three well matched groups of subjects: non-diabetic controls, diabetics without retinopathy, and diabetics with proliferative retinopathy. /8-thromboglobulin and platelet factor 4 concentrations were significantly higher in the diabetics with retinopathy than in the controls and platelet factor 4 was also increased in the diabetics without retinopathy compared with controls. Fibrinogen concentration was raised in diabetics without retinopathy compared with controls, diabetics with retinopathy compared with controls, and diabetics with retinopathy compared with those without. Fibrinopeptide A concentration did not differ significantly between groups. III levels were increased in diabetics with retinopathy compared with controls, and in diabetics with retinopathy compared with those without. Factor copyright. VIII related antigen values were higher in both the diabetic groups when compared with the controls. Fibrinopeptide A concentration correlated with both /8-thromboglobulin and platelet factor 4 in each of the three groups. Haemostatic abnormalities in diabetes have been shown, although a hypercoagulable state has not been confirmed. These changes in platelet and coagulation function may be secondary to the development of microvascular disease and their role in the pathogenesis of retinopathy remains http://jcp.bmj.com/ uncertain.

Patients with diabetes mellitus have significantly We have measured plasma concentrations of increased morbidity and mortality as a consequence ,8-thromboglobulin, platelet factor 4, fibrinogen, of specific microvascular disease, which results in fibrinopeptide A, antithrombin III, factor VIII conditions such as retinopathy and nephropathy. related antigen, a2-macroglobulin, platelet count, on September 28, 2021 by guest. Protected Although the precise cause of these vascular compli- and total glycosylated haemoglobin in three well cations remains uncertain, evidence is accumulating matched groups of subjects: non-diabetic controls, that an imbalance of the haemostatic mechanisms insulin dependent diabetics without retinopathy, may be entailed in their initiation or propagation. and insulin dependent diabetics with proliferative Various abnormalities of platelet function and of retinopathy. coagulation in diabetes have been identified and recently reviewed.' The contradictory findings and Patients and methods conclusions of different research groups may be par- tially attributed to the heterogeneity of their diabe- PATIENTS tic patients, the poor matching of control groups, the We studied 20 healthy non-diabetic controls, 20 non-standardisation of coagulation and platelet insulin dependent diabetics without ophthalmos- assays, and the uncertain importance of abnormal in copic evidence of retinopathy, and 20 insulin vitro tests. dependent diabetics with proliferative retinopathy. Subjects in each group were individually matched Accepted for publication 20 February 1984 for age, sex and, where applicable, duration of diab- 659 J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from 660 Borsey, Prowse, Gray, Dawes, James, Elton, Clarke Table 1 Details ofthe subjects studied rations was included on every plate. The standard Normal Diabetics was a human a2-macroglobulin reference obtained subjects from Meloy Laboratories Inc (Springfield, Virginia, without with USA). a2-macroglobulin was measured only in the retinopathy retinopathy diabetic subjects. No of subjects 20 20 20 % male 60 60 60 Glycosylated haemoglobin-measured by a mod- Age (yr) 35 ± 9 34 9 35 ± 11 ification of the short column chromatographic Duration of diabetes (yr)- 21 + 7 19 ± 7 method of Kynoch and Lehmann.9 Daily insulin dose (units)- 62 + 18 68 ± 29 % ideal body weight* - 101 + 11 100 + 11 Platelet count-electronically determined using the Thrombocounter (Coulter Ltd). Values given as mean ± 1 SD. *Metropolitan Life Assurance Company Tables 1960. The assay methods for 3-thromboglobulin, platelet factor 4, fibrinopeptide A, antithrombin III activity, antithrombin III antigen, and factor VIII etes. Details of the subjects studied are given in related antigen gave within batch coefficients of var- Table 1. All diabetics had presented with the clinical iation between 4 and 6% with between batch features of type I diabetes before the age of 35 and coefficients of variation of 8-12%. The between been treated with insulin since diagnosis. batch coefficient of variation for glycosylated All subjects had normal serum creatinine con- haemoglobin was 3 5%. Fibrinogen levels and centrations and absence of proteinuria on testing platelet count were determined by non-batch with Albustix. None had chronic disease or was methods with coefficients of variation of 5 % and 3% receiving drugs (for example, oral contraceptives, respectively. Previous studies by one of us have salicylates) known to interfere with haemostasis. shown that the assay procedure for a2- Only two of the diabetics studied had signs or symp- macroglobulin is highly reproducible (100 + toms, or both, of macrovascular disease (one angina 22 %).8 Further details of the above assays are and the other peripheral vascular disease) and they available in the quoted references. both had retinopathy. Laboratory staff were unaware of the identities ofcopyright. the samples. METHODS All diabetic fundi were examined ophthalmoscopi- STATISTICAL METHODS cally through dilated pupils by one observer (BFC). Non-parametric methods were used to analyse the Proliferative retinopathy was defined as the pres- data in view of the presence of skew and outliers in ence of neovascularisation. A single, 2 h postpran- at least some of the measurements. Median and dial venous blood sample was withdrawn into a range were used as summary statistics within groups, http://jcp.bmj.com/ polypropylene syringe, with minimal venous stasis, paired measurements were tested between groups from resting subjects. The samples were analysed as by Wilcoxon's signed rank tests, and associations detailed below: between measurements within groups by Kendall /3-thromboglobulin-radioimmunoassay of samples rank correlation coefficients. collected into mixture containing prostaglandin E,.23 Results Platelet factor 4-radioimmunoassay.4 on September 28, 2021 by guest. Protected Fibrinopeptide A-radioimmunoassay of samples Table 2 details the medians and ranges of the meas- collected into -heparin.5 ured platelet and coagulation factors within the Fibrinogen (factor 1)-clot turbidity assay of Ellis three groups and also the results of a comparison of and Stransky.6 paired measurements between groups. Antithrombin III /3-thromboglobulin concentrations were higher (p (a) Antithrombin III activity-bioassay with < 0.01) in the diabetics with retinopathy than in the chromogenic substrate using a commercially avail- non-diabetic controls, and the platelet factor 4 val- able kit (Boehringer-Mannheim Ltd). ues were significantly increased in both the diabetic (b) Antithrombin III antigen-electroimmunoassay groups when compared with the controls (diabetics method of Laurell using commercial antisera with retinopathy p < 0-01; diabetics without (Behringwerke).' retinopathy p < 0.05). A highly significant correla- Factor VIII related antigen-electroimmunoassay tion between /8-thromboglobulin and platelet factor method of Laurell using commercial antisera 4 were seen in each of the three groups (controls and (Behringwerke).' diabetics with retinopathy p < 0.001; diabetics ar2-macroglobulin-standard Mancini gel diffusion without retinopathy p < 0-01). method.8 Standard antigen in four different concent- The fibrinogen concentrations were higher in the J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from Platelet and coagulation factors in proliferative diabetic retinopathy 661 Table 2 Platelet and coagulation factors (median and range) in the three study groups and a comparison ofpaired measurements between groups Non-diabetic Diabetics Controls v Controls v Diabetics with controls diabetics without diabetics with retinopathy v those without retinopathy with retinopathy retinopathy retinopathy without ,S-thromboglobulin (ng/ml) 31-5 (19-1-68-5) 44-9 (22-2-96-3) 51-5 (24-1-127-5) NS t NS Platelet factor 4 (n$/rni) 10-4 (10.0-52-0) 17-5 (8.5-72.6) 19-6 (6-7-63.7) * t NS Fibnnogen (gil) 1-70 (1-04-2-93) 2-46 (1-56-3-46) 2-93 (1-85-5-70) t * Fibrinopeptide A (nglml) 1-64 (1-60-19-20) 2-82 (2-00-16,60) 2-80 (2-00-17-80) NS NS NS Antithrombin III activity (%) 103 (82-119) 104 (92-122) 116 (92-132) NS t Antithrombin HI antigen (%) 100 (76-126) 100 (87-121) 113 (87-126) NS * * Factor VW1 related antigen (%) 98 (72-140) 252 (126-460) 267 (142-400) t NS a2macroglobulin (mc%o - 378 (192-641) 480 (187-696) - - NS Platelet count (x109/l) 226 (106-284) 255 (104-420) 267 (126-350) * NS NS Glycosylated haemoglobin (%) 6-6 (5 6-8 3) 10-7 (8-3-14-1) 11-7 (10-2-15-9) t * NS = not significant. *p < 005, fp < 001, tp < 0-001. two diabetic groups than in the controls and also in related significantly with /3-thromboglobulin (p < the diabetics with retinopathy than in those without 0.05) and platelet factor 4 (controls and diabetics retinopathy. without retinopathy p < 0-05; diabetics with There was no significant difference in fibrinopep- retinopathy p < 0.01) (Figure). tide A concentrations between the three groups. Antithrombin III (activity and antigen) levels copyright. Within each group, however, fibrinopeptide A cor- were significantly higher in the diabetics with retinopathy than in either of the other groups. There was a strong correlation between antithrombin III 20- activity and antigen in each of the three groups (con- 0 trols p < 0-01; diabetic groups p < 0-001). The

18- concentration of a2-macroglobulin was higher in the http://jcp.bmj.com/ 0 diabetics with retinopathy, than in those without but A 1 this difference was not significant. a2-macroglobulin - A correlated with antithrombin III antigen only in the A 14- 0 diabetics with retinopathy. There was no correlation between a2-macroglobulin and antithrombin activ- ity. A Concentrations of factor VIII related antigen 10- on September 28, 2021 by guest. Protected 0 were higher in both the diabetic groups than in the A non-diabetic controls (p < 0.001). L, 8- The platelet count was higher in the non-

4 0 retinopathic diabetics than in the control subjects (p 6- 0 < 0.05), but no other significant differences existed al Ao between groups.

AA 0 Normals Total glycosylated haemoglobin was higher in the 14 l 9 A No DR diabetics with retinopathy than in those without (p 10 3 ° DR < 0.05). The relation between glycosylated haemo- globin and the haemostatic factors studied was consi- I I I I I I 0 10 20 30 40 50 60 70 80 dered in each of the three groups: the only Plasma platelet factor 4 (ngJml) significant correlations were with fibrinogen in the Correlation ofplatelet factor 4 with fibrinopeptide A. control group (p < 0-001), with platelet factor 4 in < Normals: r = 0-45, p < 0 05; no diabetic retinopathy (No the uncomplicated diabetics (p 0-05), and with DR): r = 0-36, p < 0-05; diabetic retinopathy (DR): r = fibrinopeptide A and a2-macroglobulin in the diabe- 0*44, p < 0.05. tics with retinopathy (p < 0-05). J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from 662 Borsey, Prowse, Gray, Dawes, James, Elton, Clarke Discussion presence of renal impairment.420 The platelet factor 4 results were, as expected, The published work provides a confusing picture of similar to those for 13-thromboglobulin, although the importance of haemostatic abnormalities in the both diabetic groups had significantly higher levels development of diabetic retinopathy. The poor than the non-diabetics. Platelet factor 4 concentra- matching of study groups, the non-standardisation tion has been previously reported as both normal'7 22 of assays, the uncertain relevance of in vitro tests to and increased in diabetics.'5 23 in vivo function, and the inclusion of patients with Fibrinopeptide A, which is cleaved from coincident nephropathy and macroangiopathy have fibrinogen by , is a sensitive indicator of all contributed to this confusion. The subjects in our thrombin formation24 and the lack of a significant groups were individually matched for age, sex, and, difference in concentrations between our three where applicable, duration of diabetes, and all groups is evidence against activation of the coagula- diabetics with nephropathy were excluded. In con- tion system in diabetes. Our finding that fibrinopep- trast to many previous studies, our two diabetic tide A concentrations correlated with both groups were chosen to represent the extremes of the f3-thromboglobulin and platelet factor 4 concentra- " retinopathy spectrum"-namely, those without tions suggests that thrombin may initiate the platelet retinopathy and those with proliferative release reaction. A close association between retinopathy. Unfortunately, the exclusion of thrombin formation and the platelet release reaction patients with asymptomatic large vessel disease in normal subjects has been previously noted in remains a problem since it necessitates unacceptable vitro25 and a correlation between fibrinopeptide A extensive and invasive investigation. and /8-thromboglobulin in diabetics has been The platelet specific f8-thromboglobulin reported by Burrows.2' and platelet factor 4 are released together from a Fibrinogen is the immediate precursor of , granules during platelet aggregation, and increased influences plasma viscosity,26 and is a necessary plasma levels may reflect enhanced in vivo platelet cofactor for ADP and epinephrine induced platelet activation. Their precise biological functions remain aggregation.27 It is generally agreed that fibrinogen uncertain. f8-thromboglobulin may reduce pros- concentrations are increased in diabetic patientscopyright. tacyclin generation in the vessel wall'0 and thus when compared with controls28-30 and several allow a more sustained platelet-endothelial interac- authors have, as in our study, shown higher concent- tion. Platelet factor 4 has heparin neutralising activ- rations in diabetics with retinopathy than in those ity" and may neutralise any natural heparin like without.29-31 This increase in fibrinogen concentra- inhibitors of coagulation. It has been suggested that tion may be non-specific, perhaps in response to

intensive release of platelet factor 4 modifies the chronic vascular injury, rather than reflecting a http://jcp.bmj.com/ surface properties of the platelet membrane and hypercoagulable state.' The higher concentrations thereby accelerates the rate of formation of the of other acute phase reactants (such as haptoglobin, fibrin network between .'2 A close correla- ,8-lipoprotein, and caeruloplasmin) in the serum of tion between the plasma levels of platelet factor 4 diabetic subjects32 lends support to this view. Our and /3-thromboglobulin has been well study has failed to confirm a reported correlation described.4 " '5 between glycosylated haemoglobin and fibrinogen in Both our diabetic groups had higher diabetics.33 ,3-thromboglobulin concentrations than the con- Antithrombin III is the major inhibitor of the on September 28, 2021 by guest. Protected trols, although this was significant only in patients activated coagulant enzymes XIIa, XIa, IXa, Xa, with retinopathy. Both normal'6 7 and and thrombin. The risk of thrombo-embolic disease increased2 1518-20 plasma f8-thromboglobulin con- in patients with congenital antithrombin III defi- centrations have been reported in diabetes. The lack ciency is significantly increased.34 In a hypercoagul- of a significant difference in ,8-thromboglobulin con- able state antithrombin III would be consumed and centrations between the diabetics with retinopathy low circulating levels expected. Our findings of and those without is in keeping with a previous raised antithrombin III levels in those diabetics with report from Edinburgh,2 but contrasts with work retinopathy was therefore somewhat surprising. from other centres.'5 18-21 The inconsistent findings There have been reports of low,35-37 normal,3839 and of previous studies may be partially attributed to the raised222940 levels of antithrombin III in diabetes. use of different anticoagulant mixtures, sampling Discrepancies between functional and immunologi- techniques, and assay methods2 and to the inclusion cal assays may account for some of these inconsis- of diabetics with nephropathy. The latter is impor- tencies, although we found a strong correlation bet- tant since plasma 83-thromboglobulin (but not ween antithrombin III activity and antigen. What platelet factor 4) concentrations are increased in the then is the explanation of our findings? The raised J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from Platelet and coagulation factors in proliferative diabetic retinopathy 663 levels of antithrombin III may simply represent part eral of our findings may be explained on the basis of of an acute phase reaction occurring subsequent to a non-specific stress reaction in diabetics, possibly the development of retinopathy. Alternatively, an secondary to chronic vascular damage. Further increase in antithrombin III may occur as a defence studies, particularly of a prospective nature, are response to enhanced intravascular production of clearly indicated to determine whether the observed thrombin via a feedback mechanism increasing changes are causal or secondary to the vascular synthesis or decreasing catabolism of antithrombin complications of diabetes mellitus. III.4' a2-macroglobulin is a further inhibitor of coagula- tion proteases which accounts for about a quarter of the total functional antithrombin III activity in the References blood. Indeed, Elder et a140 attributed the raised Jones RL, Peterson CM. Haematologic alterations in diabetes levels of antithrombin III activity in their diabetics mellitus. Am J Med 1981;70:339-52. to an increase in a2-macroglobulin concentrations. 2 Borsey DQ, Dawes J, Fraser DM, Prowse CV, Elton RA, Clarke BF. Plasma beta-thromboglobulin in diabetes mellitus. We found no correlation between antithrombin Diabetologia 1980;18:353-7. III activity and a2-macroglobulin. Although it is 3 Bolton AE, Ludlam CA, Moore S, Pepper DS, Cash JD. Three generally accepted that a2-macroglobulin levels are approaches to the radioimmunoassay of human higher in diabetics,3' 4243 the relation between 3-thromboglobulin. Br J Haematol 1976;33:233-8. Bolton AE, Ludlam CA, Pepper DS, Moore S, Cash JD. A retinopathy and a2-macroglobulin has remained radioimmunoassay for platelet factor 4. Thromb Res uncertain because of other variables including age, 1976;8:51-8. duration of diabetes, metabolic control and neph- Kockum C. Radioimmunoassay of fibrinopeptide A: Clinical ropathy, which may influence a2-macroglobulin applications. Thromb Res 1976;8:225-36. 6 Ellis BC, Stransky A. A quick and accurate method for the levels.4244 Previous studies, which often failed to determination of fibrinogen in plasma. J Lab Clin Med consider these factors, have reported both higher 1961;58:477-88. a2-macroglobulin concentrations in diabetics with Laurell CB. Quantitative estimation of proteins by elec- retinopathy3' 43 and comparable levels in retinopaths trophoresis in agarose gel containing antibodies. Anal

Biochem 1966; 15:45-52. copyright. and non-retinopaths.42 Our results are in keeping James K, Johnson G, Fudenberg HH. The quantitative estima- with those of Gray et al,45 who found no significant tion of a2-macroglobulin in normal, pathological and cord difference in a2-macroglobulin levels between sera. Clin Chim Acta 1966;14:207-14. diabetics with retinopathy and those without when 9Borsey DQ, Fraser DM, Gray RS, Elton RA, Smith AF, Clarke BF. Glycosylated haemoglobin and its temporal relationship to all the above variables were taken into account. plasma glucose in non-insulin dependent (Type 2) diabetes Factor VIII related antigen is a , mellitus. Metabolism 1982;31:362-6. '° Hope W, Chesterman CN, Dusting GJ, Smith I, Morgan FJ, higher oligomers of which are active in promoting http://jcp.bmj.com/ platelet adhesion. It is found only in endothelial cells Martin TJ. Inhibition by f3-thromboglobulin of prostacyclin (PGI2) formation in arterial endothelial cells. Thromb Haem- (where it is synthesised), platelets, and megakaryo- ost 1979;42:8. cytes." Von Willebrand's factor appears to be a "Van Creveld S, Paulssen MMP. Significance of clotting factors in property of factor VIII related antigen4' and these blood-platelets in normal and pathological conditions. Lancet factors are associated in plasma with factor VIII 1951;ii:242-4. Niewiarowski S, Lipinski B, Farbiszewski R, Poplowski A. The coagulant activity. Our finding of increased factor release of platelet factor 4 during platelet aggregation and the VIII related antigen levels in diabetic patients is in possible significance of this reaction in haemostasis. Experien- keeping with previous reports33 48 49 and could result tia 1968;24:343-4. on September 28, 2021 by guest. Protected from enzymic degradation in the circulation of the 3 Dawes J, Smith RC, Pepper DS. The release, distribution and clearance of human B-thromboglobulin and platelet factor 4. factor VIII polymer and/or from greater endothelial Thromb Res 1978;12:851-61. synthesis and release of factor VIII related anti- 4 Kurihara Y, Nakayama H, Nakagawa S. Plasma gen.51 ' The lack of raised plasma fibrinopeptide A ,B-thromboglobulin and platelet factor 4 in renal insufficiency. levels supports the latter explanation since in an Throm Res 1980;18:557-60. "Betteridge DJ, Zahavi J, Jones NAG, Shine B, Kakkar VV, Gal- enzymatically active environment one would expect ton DJ. Platelet function in diabetes mellitus in relationship to increased thrombin formation. We found no complications, glycosylated haemoglobin and serum lipopro- significant difference in factor VIII related antigen teins. Eur J Clin Invest 1981;11:273-7. values between diabetics with retinopathy and 6 Campbell IW, Dawes J, Fraser DM, et al. Plasma ,8-thromboglobulin in diabetes mellitus. Diabetologia those without, although higher levels have been pre- 1977;26: 1175-7. viously reported in diabetics with vascular dis- 7 Kazmier FJ, Bowie EJ, O'Fallon WM, Zimmerman BR, ease. 48 50 52 Osmundson PJ, Palumbo PJ. A prospective study of peripheral This study has shown certain platelet and coagula- occlusive arterial disease in diabetes IV: Platelet and plasma functions. Mayo Clin Proc 1981;56:243-53. tion abnormalities in diabetes, but the presence of a 18 Preston FE, Ward JD, Marcola BH, Porter NR, Timperley WR, hypercoagulable state has not been confirmed. Sev- O' Malley BC. Elevated 1-thromboglobulin levels and circulat- J Clin Pathol: first published as 10.1136/jcp.37.6.659 on 1 June 1984. Downloaded from 664 Borsey, Prowse, Gray, Dawes, James, Elton, Clarke ing platelet aggregates in diabetic microangiopathy. Lancet 36 Monnier L, Follea G, Mirouze J. Antithrombin III deficiency in 1978;1:238-9. diabetes mellitus: influence on vascular degenerative compli- '9 Burrows AW, Chavin SI, Hockaday TDR. Plasma- cations. Hormn Metab Res 1978;10:470-3. thromboglobulin concentrations in diabetes mellitus. Lancet 37 Sowers JR, Tuck ML, Sowers DK. Plasma antithrombin Ill and 1978;1:235-7. thrombin generation time: correlation with haemoglobin A, 20 Rasi V, Ikkala E, Hekali R, Myllyla G. Factors affecting plasma and fasting serum glucose in young diabetic women. Diabetes ,B-thromboglobulin in diabetes mellitus. Med Biol Care 1980;3:655-8. 1980;58:269-72. 38 Zucker ML, Gomperts ED, Russel D, et at. Antithrombin func- 21 Burrows AW. Beta-thromboglobulin in diabetes: relationships tional activity after saturated and unsaturated fatty meals and with blood glucose and fibrinopeptide A. Horm Metab Res fasting in normal subjects and some disease states. Thromb 1981; suppl II: 22-5. Res 1979;15:37-48. 22 Corbella E, Miragliotta G, Masperi R, Villa S, Bini A, de 39 Gandolfo GM, DeAngelis A, Torresi MV. Determination of Gaetano G, Chiumello G. Platelet aggregation and antithrom- antithrombin III activities by different methods in diabetic bin III levels in diabetic children. Haemostasis 1979;8:30-7. patients. Haemostasis 1980;9:15-9. 23 Chimielewski J, Farbiszewski R. Platelet factor 4 (PF4) release 40 Elder GE, Mayne EE, Daly JG, Kennedy AL, et al. Antithrom- during human platelet aggregation in diabetic patients. bin III activity and other coagulation changes in proliferative Thromb Diath Haemorrh 1970;24:203-5. diabetic retinopathy. Haemostasis 1980;9:288-96. 24 Nossel HL, Ti M, Kaplan KL, Spanondis K, Soland T, Butler VP. "1 Calvalho A, Ellman L. Activation of the coagulation system in The generation of fibrinopeptide A in clinical blood samples: polycythemia vera. Blood 1976;47:669-78. evidence for thrombin activity.J Clin Invest 1976;58: 1136-44. 42 Ganrot PO, Gydell K, Ekelund H. Serum concentration of 2S Prowse CV, Vigano S, Borsey DQ, Dawes J. The release of alpha2-macroglobulin, haptoglobin and alpha2-antitrypsin in beta-thromboglobulin from platelets during the clotting of diabetes mellitus. Acta Endocrinol 1967;55:537-44. whole blood. Thromb Res 1980;17:433-42. 43 James K, Merriman J, Gray RS, Duncan LJP, Herd R. Serum 26 Hoare EM, Barnes AJ, Dormandy JA. Abnormal blood viscosity a2-macroglobulin levels in diabetes. J Clin Pathol in diabetes mellitus and retinopathy. Biorheology 1980;33:163-6. 1976;13:21-5. "Ganrot PO, Schersten B. Serum a2-macroglobulin concentration 27 Solum NO, Stormorken H. Influence of fibrinogen on the and its variation with age and sex. Clin Chim Acta aggregation of washed human blood platelets induced by 1967;15:113-20. adenosine diphosphate, thrombin, collagen and adrenaline. Gray RS, James K, Merriman J et al. Alpha2-macroglobulin and Scand J Clin Lab Invest 1965;17 (suppl 84): 170-82. proliferative retinopathy in Types I Diabetes. Horm Metab Res 28 Mayne JM, Bridges JM, Weaver JA. Platelet, plasma fibrinogen 1982;14:389-92. and factor VIII levels in diabetes mellitus. Diabetologia 46Jaffe EA, Deykin D, Salzman E, Schnippe L. Endothelial cells

1970;6:436-40. and biology of Factor VIII. N Engl J Med 1977;296:377-83.copyright. 29 Fuller H, Keen JH, Jarrett RJ, et al. Haemostatic variables 47 Von Willebrand's disease (Editorial). Br Med J 1976;ii:715-6. associated with diabetes and its complications. Br Med J 48 Pandolfi M, Almer LO, Holmberg L. Increased von 1978;ii:964-6. Willebrand-antihaemophilic factor A in diabetic retinopathy. 30 Cederholm-Williams SA, Dornan TL, Turner RC. The metabol- Acta Ophthalmol 1974;52:823-8. ism of fibrinogen and plasminogen related to diabetic 49Gensini GF, Abbate R, Favilla S, Neri Semeri GG. Changes of retinopathy in man. Eur J Clin Invest 1981;11:133-8. platelet function and blood clotting in diabetes mellitus. 31 Almer LA, Pandolfi M. and diabetic retinopathy. Thromb Haemost 1979;42:983-93. Diabetes 1976;25 (suppl 2): 807-9. 50 Abbal Plante Bierme R. Factor VIII and Boneu B, M, J, complex http://jcp.bmj.com/ 32 Jonsson A, Wales ZK. Blood glycoprotein levels in diabetes mel- damage. Lancet 1975;i: 1430. litus. Diabetologia 1976;12:245-50. -1 Giustolisi R, Musso R, Russo M, Cantania N, Lombardo T, Cac- 33 Coller BS, Frank RN, Milton RC, Gralnick HR. Plasma co- ciola E. Possible evidence for increased Factor VIII antigen factors of platelet function: correlation with diabetic synthesis in vascular endoth#lium of diabetic subjects. Thromb retinopathy and haemoglobins Ala-c. Ann Intern Med Haemost 1982;47:293. 1978;88:311-6. 52 Lufkin EG, Fass DN, O'Fallon WM, Bowie EJW. Increased von 34 Marciniak E, Farley CH, DeSimone PA. Familial due Willebrand factor in diabetes mellitus. Metabolism to antithrombin III deficiency. Blood 1974;43:219-31. 1979;28:63-6. 3S Banerjee RN, Sahni AL, Kumere U, Arza N. Antithrombin III deficiency in maturity onset diabetes mellitus and atherosc- Requests for reprints to: Dr DQ Borsey, Ninewells Hospi- on September 28, 2021 by guest. Protected lerosis. Thromnb Diath Haemorrh 1974;31:339-45. tal, Ninewells, Dundee DD1 9SY, Scotland.