Relative risk (RR) of 5 or more microaneurysms(MAs) in those randomized to the tight (TBP) and less tight blood pressure(LTBP) control groups. A, Overall randomization. B, Those with no retinopathyat randomization. C, Those with any type of retinopathy at randomization.CI indicates confidence interval.
Relative risk (RR) of hard exudates(HEs) in those randomized to the tight (TBP) and less tight blood pressure(LTBP) control groups. A, Overall randomization. B, Those with no retinopathyat randomization. C, Those with any type of retinopathy at randomization.CI indicates confidence interval.
Relative risk (RR) of at least 1cotton-wool spot (CWS) in those randomized to the tight (TBP) and less tightblood pressure (LTBP) control groups. A, Overall randomization. B, Those withno retinopathy at randomization. C, Those with any type of retinopathy atrandomization. CI indicates confidence interval.
Relative risk (RR) of 2-step or worsedeterioration on the Early Treatment Diabetic Retinopathy Study (ETDRS) scalein those randomized to the tight (TBP) and less tight blood pressure (LTBP)control groups. A, Overall randomization. B, Those with no retinopathy atrandomization. C, Those with any type of retinopathy at randomization. CIindicates confidence interval.
Relative risk (RR) of retinal photocoagulationin those randomized to the tight (TBP) and less tight blood pressure (LTBP)control groups. CI indicates confidence interval.
Relative risk (RR) of 3 lines ormore of deterioration in either eye in those randomized to the tight (TBP)and less tight blood pressure (LTBP) control groups. CI indicates confidenceinterval.
UK Prospective Diabetes Study (UKPDS) Group. Risks of Progression of Retinopathy and Vision Loss Related to TightBlood Pressure Control in Type 2 Diabetes MellitusUKPDS 69. Arch Ophthalmol. 2004;122(11):1631-1640. doi:10.1001/archopht.122.11.1631
To determine the relationship between tight blood pressure (BP) controland the different aspects of diabetic retinopathy in patients with type 2diabetes mellitus (DM).
Nineteen hospital-based clinics in England, Scotland, and Northern Ireland.
Outcome of retinopathy status assessed by 4-field retinal photographyrelated to allocation within a randomized controlled trial comparing a tightBP control policy aiming for a BP less than 150/85 mm Hg with a less tightBP control policy aiming for a BP less than 180/105 mm Hg.
One thousand one hundred forty-eight hypertensive patients with type2 DM were studied. These patients had type 2 DM for a mean duration of 2.6years at the inception of the Hypertension in Diabetes Study, had a mean ageof 56 years; and had a mean BP of 160/94 mm Hg. Seven hundred fifty-eightpatients were allocated to a tight BP control policy with angiotensin-convertingenzyme inhibitor or β-blockers as the main therapy; 390 were allocatedto a less tight BP control policy.
Main Outcome Measures
Deterioration of retinopathy (≥2-step change on a modified EarlyTreatment Diabetic Retinopathy Study [ETDRS] final scale), together with endpoints (photocoagulation, vitreous hemorrhage, and cataract extraction) andanalysis of specific lesions (microaneurysms, hard exudates, and cotton-woolspots). Visual acuity was assessed at 3-year intervals using ETDRS logarithmof the minimum angle of resolution charts. Blindness was monitored as an endpoint with the criterion of Snellen chart assessment at 6/60 or worse.
By 4.5 years after randomization, there was a highly significant differencein microaneurysm count with 23.3% in the tight BP control group and 33.5%in the less tight BP control group having 5 or more microaneurysms (relativerisk [RR], 0.70; P = .003). The effectcontinued to 7.5 years (RR, 0.66; P<.001). Hardexudates increased from a prevalence of 11.2% to 18.3% at 7.5 years afterrandomization with fewer lesions found in the tight BP control group (RR,0.53; P<.001). Cotton-wool spots increased inboth groups but less so in the tight BP control group which had fewer cotton-woolspots at 7.5 years (RR, 0.53; P<.001). A 2-stepor more deterioration on the ETDRS scale was significantly different at 4.5years with fewer people in the tight BP control group progressing 2 stepsor more (RR, 0.75; P = .02). Patients allocatedto tight BP control were less likely to undergo photocoagulation (RR, 0.65; P = .03). This difference was driven by a differencein photocoagulation due to maculopathy (RR, 0.58; P = .02).The cumulative incidence of the end point of blindness (Snellen visual acuity,≥6/60) in 1 eye was 18/758 for the tight BP control group compared with12/390 for less tight BP control group. These equate to absolute risks of3.1 to 4.1 per 1000 patient-years, respectively (P = .046;RR, 0.76; 99% confidence interval, 0.29-1.99). There was no detectable differencein outcome between the 2 randomized therapies of angiotensin-converting enzymeinhibition and β-blockade.
High BP is detrimental to each aspect of diabetic retinopathy; a tightBP control policy reduces the risk of clinical complications from diabeticeye disease.
Type 2 diabetes mellitus (DM) and hypertension are frequently associated,often within the context of the metabolic syndrome, where obesity and dyslipidemiaare prominent. The prevalence of hypertension in type 2 DM may be higher thanin the general population. At age 40 years, approximately 32% of the patientswith type 2 DM are hypertensive, the proportion increasing to 47% by age 60years.1 Hypertension increases risk for thedevelopment of microvascular disease and the UK Prospective Diabetes Study(UKPDS) Group has documented both the prevalence and the extent to which interventionto reduce blood pressure (BP) reduced the incidence of microvascular end points.2 The reduction of systolic BP (SBP) by a median 10mm Hg with diastolic BP (DBP) reduction of 5 mm Hg resulted in a 37% decreasein microvascular disease, and an observational analysis has demonstrated theabsence of any threshold of hypertension effect between the SBP limits of120 and 180 mm Hg.3 Microvascular disease,for the previous analyses, was a composite of predefined conditions that includedphotocoagulation, vitreous hemorrhage, and renal disease (plasma creatininelevel, >2.8 mg/dL [>250 μmol/L]). We present the specific retinopathy data(distinct from the wider-defined microvascular disease) as assessed by 4-fieldfundal photography, eye-related end points, and visual acuity.
Seven thousand six hundred sixteen patients with newly diagnosed type2 DM were referred to the 23 hospital recruiting clinics within the UnitedKingdom between December 8, 1977, and December 31, 1991. Exclusion criteriahave previously been published.4 Of the patientsreferred, 5102, who had fasting plasma glucose values greater than 108 mg/dL(>6.0 mmol/L) on 2 separate mornings and who fulfilled all other inclusioncriteria, were recruited. Those with malignant hypertension and those withpreexisting retinopathy needing laser treatment were excluded. The study conformedto the guidelines of the Declarations of Helsinki (1975 and 1983). All patientsgave informed consent. The study received institutional review board approvalfrom the Central Oxford Ethics Committee, Oxford, England.
In 1987, a BP control study was introduced in a factorial manner withthe glucose control study.4 One thousand onehundred forty-eight hypertensive patients with type 2 DM were studied. Thesepatients had type 2 DM for a mean duration of 2.6 years and an SBP greaterthan 160 mm Hg and/or a DBP greater than 90 mm Hg (if not receiving treatmentfor hypertension) or an SBP greater than 150 mm Hg and/or a DBP greater than85 mm Hg (if already receiving treatment for hypertension) were randomly allocatedbetween a less tight BP control policy, aiming for an SBP less than 180 mmHg and a DBP less than 105 mm Hg, and a tight BP control policy, with a randomallocation to either an angiotensin-converting enzyme (ACE) inhibitor or a β-blocker,aiming for an SBP less than 150 and a DBP less than 85 mm Hg.
Within the UKPDS glucose control study, patients were treated by dietalone for 3 months. Patients who remained hyperglycemic (fasting plasma glucoselevel, 110-270 mg/dL [6.1-15.0 mmol/L]) in the absence of diabetic symptomswere randomized to a conventional blood glucose control policy, primarilywith diet or to an intensive policy (aiming for a fasting plasma glucose level<108 mg/dL [<6.0 mmol/L]) with either additional sulfonylurea, insulin,or metformin therapy.
Of 20 UKPDS centers included in the Hypertension in Diabetes Study,19 took retinal photographs. Of the 4297 patients recruited, 243 had eitherdied or were lost to follow-up prior to its start in 1987. Of the remaining4054 patients, 1544 (38%) had hypertension, defined in patients not receivingantihypertensive therapy who had an SBP of 160 mm Hg or higher and/or a DBPof 90 mm Hg or higher or in patients receiving antihypertensive therapy asan SBP of 150 mm Hg or higher and/or a DBP of 85 mm Hg or higher. Patientswere enrolled on the basis of the mean of 3 BP measurements taken at consecutiveclinic visits. Of the 1544 hypertensive patients, 252 were excluded and 144patients did not enroll in the study. A total of 1148 patients (54% male),mean (SD) age of 56.4 (8.1) years were randomized; 727 had no previous therapyand 421 were previously treated for hypertension.
Two thirds of the patients (n = 758) were randomized to atight BP control policy, aiming for BP less than 150/85 mm Hg (with 400 patientsallocated to an ACE inhibitor, captopril, and 358 to a β-blocker, atenolol,as the main therapy) and one third of the patients (n = 390) wererandomized to a less tight BP control policy, aiming for BP of 180/105 mmHg or less but avoiding therapy with ACE inhibitors or β-blockers. Therandomization was stratified for those with or without previous therapy forhypertension. The original allowable upper limit of 200/105 mm Hg in the lesstight BP control group was reduced to 185/105 mm Hg in 1992 by the steeringcommittee following publication of the results of studies of elderly, nondiabeticsubjects in the years 1991-1992.5- 7
Captopril therapy was usually started at a dose of 25 mg twice dailyincreasing to 50 mg twice daily. Atenolol therapy was usually started at adaily dose of 50 mg increasing to 100 mg, if required. If control criteriawere not met in the tight BP control group despite maximum allocated therapyor, in the less tight BP control group without drug therapy, other agentswere added, the suggested sequence being frusemide, 20 mg daily (maximum 40mg twice daily); slow-release nifedipine, 10 mg (maximum 40 mg) twice daily;methyldopa, 250 mg (maximum 500 mg) twice daily; and prazosin, 1 mg (maximum5 mg) thrice daily. Patients were seen at 3- to 4-monthly clinic visits.
Sitting BP (diastolic phase 5) was measured by a trained nurse, afterat least 5-minutes rest, with an electronic, auscultatory blood pressure readingmachine (Copal UA-251 or a Takeda UA-751; Andrew Stephens Co, Brighouse, England)or with a random zero sphygmomanometer (Hawksley & Sons Ltd, Sussex, England)in patients with atrial fibrillation. The first reading was discarded andthe mean of the next 3 consecutive readings with a coefficient of variationbelow 15% was used in the study, with additional readings, if required. Monthlyquality assurance measurements showed the mean (SD) difference between theTakeda and Hawksley machines to be 1(4) mm Hg or less. Doppler BP readingswere taken every 3 years.
At enrollment to the UKPDS and subsequently every 3 years thereafter,patients underwent a clinical examination that included retinal color photography,ophthalmoscopy by a diabetologist clinician or an ophthalmologist, and recordingof visual acuity. Annual direct ophthalmoscopy was also carried out and achecklist for clinical events completed. Visual acuity was measured usingSnellen charts until 1989 and subsequently with ETDRS logarithm of the minimumangle of resolution (logMAR) charts with best-corrected vision, current refraction,or through a pinhole. Retinal color photographs of 4 standard 30° fieldsper eye (temporal to macular, macular, disc, and nasal fields) were takenin duplicate or with stereopsis, with additional stereophotographs of themacula. A second photograph was taken if the quality of the photograph wasunsatisfactory. Retinal photographs were masked to avoid any patient identificationprior to being assessed at a central grading center. Assessment involved aninitial review by 2 independent assessors for image quality, adherence toprotocol, and the presence or absence of diabetic retinopathy. Any eyes withretinopathy were then graded by 2 independent senior assessors (one of whomwas S.J.A.). Retinopathy lesions were assessed against corresponding EarlyTreatment Diabetic Retinopathy Study (ETDRS) standard photographs or measurements,8 following which a computerized algorithm allocateda retinopathy severity score to the eye using a modified version of the ETDRSfinal scale. The ETDRS final scale, together with a description of their clinicalfeatures, is given in Table 1. Differencesin opinion between assessors at any stage were managed by independent adjudication.The numeric scale was then used to derive a worse eye/better eye score.8 Microaneurysms (MAs) were counted in each eye, avoidingoverlapping fields, and summated.
Randomization into the hypertension study was not tied to annual ortriennial visits of the UKPDS. To form the baseline data set, we used theretinal photograph taken up to 3 years prior to hypertension randomization.We then report intervals of 1.5, 4.5, and 7.5 years—the median intervalfrom randomization for the subsequent photographs.
Retinopathy requiring photocoagulation or vitreous hemorrhage were independentlyassessed and recorded throughout the study. These data were used to augmentthe photographic evidence. The reasons for visual loss were not prospectivelycollected in the UKPDS data set and were assessed from ophthalmic notes, whereavailable, retrospectively.
All analyses were calculated on an intention-to-treat basis, comparingpatients allocated to tight or less tight BP control policy. Change in diabeticretinopathy was defined in the protocol as a 2-step or greater change in ETDRSgrading (both eyes 1 step or 1 eye ≥2 steps) with a worse eye/better eyescale including retinal photocoagulation or vitreous hemorrhage as the mostserious grade.9 We present a 3-step changefor purposes of comparability with the Diabetes Control and ComplicationsTrial.10 Visual loss was defined as the bestvision in either eye, deteriorating by 3 lines on the ETDRS chart (Table 2).
Survival function estimates were calculated using the product-limit(Kaplan-Meier) method with log rank tests and hazard ratios (relative risks[RRs]) were obtained from Cox proportional hazards models. All statisticallysignificance tests were 2-sided; 99% confidence intervals (CIs) are used forassessment of surrogate end points that were measured at triennial visits.Since these visits were synchronized with the fasting plasma glucose controlstudy and not the BP study, the results were grouped in 3-year intervals andexpressed as 1.5, 4.5, and 7.5 years from randomization. Mean (SD), geometricmean (1-SD interval), or median (IQ range) have been quoted for the biometricand biochemical variables, with Wilcoxon, t, or χ2 tests for comparison tests. The overall values for BP during a periodwere assessed for each patient as the mean during that period and for eachallocation as the mean of patients with data in the allocation. Blood pressurecontrol was assessed in the cohort of patients allocated to tight and lesstight BP control policies who had data at 9 years' follow-up.
The median time from randomization date to the end of the trial was9.3 years. The median follow-up to death, the last known date at which vitalstatus was known, or to the end of the trial was 8.4 years. The vital statuswas known at the end of the trial in all patients except 14 patients (1%)who had emigrated and a further 33 patients (3%) who could not be contactedin the last year of the study for assessment of clinical end point status.
The mean (SD) BP in the 1148 patients at randomization to tight andless tight BP control groups was similar.2 Themean (SD) cohort BP during the study over 9 years follow-up was 144 (14)/82(7) mm Hg (n = 297) for the tight BP control group and 154 (16)/87(7) mm Hg (n = 156) for the less tight BP group (each P<.001). The mean (98% CI) SBP and DBP differences were 10 (9-12)mm Hg for the tight BP control group and 5 (4-6) mm Hg for the less tightBP group.
The mean glycosylated hemoglobin level A1c over years 1 through4 was 7.2% in both groups and over years 5 through 8 was 8.3% and 8.2% inthe tight and less tight BP groups, respectively.
The RRs of 5 or more MAs in total (counting lesions in both eyes) inthe tight vs less tight BP control groups are shown in Figure 1. By 4.5 years after randomization there was a highly significanteffect with 23.3% in the tight control BP group and 33.5% in the less tightBP group having met this criterion (RR, 0.70; P = .003).The effect persisted to 7.5 years (RR, 0.66; P<.001).When the data were divided into those with no lesions at randomization (primaryprevention, Figure 2B) and those withsome detectable lesions (secondary prevention, Figure 2C), the effects were still seen at 7.5 years for both groups(RR, 0.64; P = .053, and RR, 0.73; P = .046), respectively. Comparing captoprilwith atenolol therapy, there was no difference in the effect observed withinthe tightly controlled BP group. Nor was there a detectable difference inthe trend with time between the agents at 4.5 or 7.5 years.
Hard exudates increased with time in the study, from a prevalence of11.2% to 18.3% at 7.5 years after randomization. There were significant differencesbetween the tight and less tight BP control groups, with fewer lesions foundin the tight control BP group (RR, 0.59; P<.002and RR, 0.53; P<.001 at 4.5 and 7.5 years, respectively) Figure 2A.
When data from these patients were divided into those having no lesionsat randomization (primary prevention, Figure 2B) and those having some detectable lesions (secondary prevention, Figure 2C), the effects were still seen at 7.5years for both groups. There was no difference in the observed effect withinthe tightly controlled BP group between captopril and atenolol therapy forhard exudates.
Cotton-wool spots increased throughout the trial, from an overall prevalenceof 14.0% at 1.5 years to 22.4% at 7.5 years. There was a highly significantdifference between the groups with the tight control BP group having fewerCWSs at 4.5 and 7.5 years (RR, 0.69, P = .02and RR, 0.53; P<.001, respectively) (Figure 3). These differences between the tightand less tight BP control groups were demonstrable for both primary and secondaryprevention (Figure 4B and C). An examinationof those in the tight control BP group alone, allocated to atenolol or captopriltherapy, revealed no differences between these therapies overall, nor wereany detectable in either primary or secondary prevention groups.
Two-step or more deterioration on the ETDRS scale was significantlydifferent at 4.5 years with fewer people in the tight control BP group progressing2 steps or more (RR, 0.75; P = .02) toretinopathy and was more marked at 7.5 years (RR, 0.66; P<.001, Figure 4). The effectswere similar irrespective of the retinopathy status (no retinopathy or any)at enrollment in the study (Figure 4Band C). Three-step deterioration was concordant with the 2-step changes (at4.5 years RR, 0.76; P = .06; at 7.5 yearsRR, 0.61; P<.001). More than one third of thosein the tight BP control group did not change, whereas only one fifth in theless tight BP control group remained at the same level, the differences beingattributable to worsening retinopathy. In particular, twice as many subjectsin the less tight BP control group changed by 10 steps or more.
The effect of the tight BP control group with antihypertensive treatmentcompared with less tight control BP group on the occurrence of photocoagulationis shown in Figure 5. There were manymore events relating to the occurrence of photocoagulation with maculopathythan with the development of proliferative retinopathy (78 vs 10, respectively;12 unknown).
Patients allocated to the tight BP control group were less likely toundergo photocoagulation (RR, 0.65; P = .03)(Figure 5). This difference was drivenby a difference in photocoagulation due to maculopathy (RR, 0.58; P = .02). There were no statistically significant treatmentdifferences between ACE inhibitor and β-blockade.
Five patients had a vitreous hemorrhage: 3 in the tight BP control groupand 2 in the less tight BP control group. Clearly these were too few eventsto analyze.
Thirty-six patients in the tight BP control group and 14 patients inthe less tight BP control group had cataract extractions. There was no differencein the event rates or in the incidence rate in the captopril- and atenolol-treatedgroups.
The cumulative incidence of the end point of blindness (Snellen visualacuity, ≥6/60) in 1 eye was 18/758 for the tight BP control group comparedwith 12/390 for less tight BP control group (Table 3). These equate to absolute risks of 3.1 to 4.1 per 1000patient-years, respectively (P = .046;RR, 0.76; 98% CI, 0.29-1.99). Of the recorded blindness 4 events were dueto cataract, 7 by diabetic maculopathy, 9 by other causes, and for 10 thereason was unknown. No patient became blind in both eyes. There was no detectableeffect of therapy allocation between ACE inhibitor and β-blocker.
The incidence, in either eye, of a deterioration of 0.3 on the logMARchart (approximately equivalent to 3 lines on a Snellen visual acuity chart)is shown in Figure 6 for 3, 6, and 9years (Table 4). The cumulative riskfor such a change over 9 years was RR, 0.63; (98% CI, 042-0.92, P = .002). The tight BP control group compared with the lesstight BP control group had a 47% lower risk of a deterioration in visual acuityby 3 or more lines (P = .004) on a Snellenvisual acuity chart. For 2-line deterioration we found no significant changes.
Considering visual loss (visual acuity worse than logMAR 0.3 assessedin the better eye), the proportion of patients in the tight BP control groupwas 1.7% at enrollment increasing to 8.4% at 9 years. This compared with aproportion in the less tight BP control group of 1.8% at enrollment increasingto 11.6% at 9 years.
The UKPDS groups have previously reported that intensive treatment ofthe fasting plasma glucose level and tight control of BP reduces the progressionof microvascular complications in DM.2,11- 13 Theoverall median difference in BP between the 2 intervention arms was 10 mmHg SBP and 5 mm Hg DBP. Photocoagulation used for sight-threatening retinopathy,diabetic macular edema, and proliferative retinopathy was reduced by 37% inthe tightly controlled BP group.2 The UKPDSwas the longest and most detailed clinical trial of newly diagnosed type 2DM to date. This cohort can be regarded as being representative of the UnitedKingdom as a whole, but there are caveats that apply. These include the factthat there were few nonwhites in the study, and that the eldest subjects were65 years of age at recruitment. The subjects were cared for in a routine waywith 3-month follow-up visits with physicians and nurses, and this may notalways reflect the care package elsewhere.
In this article we analyze the specific features and changes in retinopathyand ocular complications in those patients who participated in the hypertensioncontrol study. Each feature and end point of diabetic retinopathy was favorablyaffected by tight control of BP and this was further demonstrated by the aggregatemeasurement of reduction of 2- and 3-step changes in retinopathy severity,using the ETDRS grading system.
The overall magnitude of the favorable effect of BP control on retinopathyprogression is greater than has previously been described. For example inthe Wisconsin Epidemiologic Study of Diabetic Retinopathy cross-sectionalstudy, Klein et al14 found that while highBP was important, it had no effect on the incidence and progression of retinopathyover a 4-year follow-up, although DBP was higher after 10 years in those inwhom macular edema developed.15 However, thestudy by Klein et al was a cohort study rather than a clinical trial and hadno specified intervention; there was no systematic attempt to optimize BPcontrol. Furthermore, in the Klein et al study, mortality was high, so theresults are only those of the survivor population; patients who died had higherBPs than those who survived.16 In our study,we have censored data to avoid bias by survival.
Two other clinical trials have looked at BP control in subjects withtype 2 DM. In the Appropriate Blood Pressure Control in Diabetes (ABCD) Trialthe subjects were stratified on the basis of their DBP at baseline, into atrial of hypertensive subjects with a DBP of 90 mm Hg or higher (n = 470)17 and a trial of normotensive subjects (n = 480).18 Within each study subjects were randomized to eitherintensive or moderate control. In the hypertensive trial the mean BP achievedwas 132/78 mm Hg in the intensive group and 138/86 mm Hg in the moderate group.Over the follow-up period in the hypertensive trial, there was no differenceobserved in the progression of diabetic retinopathy between the intensiveand moderate control groups. However, in the normotensive trial, where themean BP in the intensive group over the follow-up period was 128/75 mm Hgin the intensive group and 137/81 in the moderate group, there was less progressionin the intensive group (34 vs 46%, P = .02).This differential effect may have been seen because hypertension seems tobe more important in initiating diabetic retinopathy than in influencing progression,19 and a higher proportion of those in the normotensiveABCD trial had no retinopathy at baseline than those in the hypertensive study.Within the Steno-2 Study20 160 subjects withtype 2 DM and microalbuminuria were randomly assigned to either conventionaltreatment in accord with national guidelines or to intensive treatment withstepwise implementation of behavior modification and pharmacologic therapythat included targets for hypertension. The intensive group achieved a significantgreater decline in SBP and DBP, and a lower risk of progression of retinopathy(hazard ratio, 0.42; 95%CI, 0.17-0.87).
The earliest clinically recognizable lesions in diabetic retinopathyare MAs, and the significance of MA counts in the risk for progression hasbeen previously reported.21 Herein we reportthat at randomization 18.8% and 18.2% of the tight BP and less tight BP controlgroups, respectively, had 5 or more MAs. By 7.5 years this increased to 29.3%and 44.8%, respectively, representing a 34% risk reduction. This is likelyto be of real clinical importance. At 7.5 years median follow-up from randomizationto tight BP control, 44% of patients had no retinopathy—that is, noteven a single MA—compared with 27% retinopathy free in the less tightlycontrolled BP group. Thus, tight BP control was delaying the onset of tissuedamage in this group, and this is concordant with the epidemiological analysisreported previously.19
In DM hard exudates often appear early in the natural history of thedisease. Hard exudates have been reported in grade 3 and grade 4 hypertensiveretinopathy, when a macular star indicates receding edema in treated hypertension.22 The proportion of patients with hard exudate 7.5years after randomization in the tight control BP group was almost half thatobserved in the less tight control BP group (14.1% vs 26.6%, respectively).Hard exudates, as a consequence of capillary leakage, would be increased bythe higher intravascular pressure in those with less tight BP control. Inthis analysis we did not specify the location of the hard exudates but theirimportance in the vicinity of the center of the macula was emphasized by theETDRS definition of clinically significant macular edema.23 Eighty-six(80%) of 108 photocoagulation episodes were undertaken because of maculopathyalone or in combination with proliferative retinopathy. There was no differencebetween the risk reduction shown in those with no or those with any retinopathyat study enrollment.
Cotton-wool spots (soft exudates) usually indicate occlusion of smallarterioles and are a well-known feature of hypertensive retinopathy. In DM,vascular occlusion occurs even in the absence of hypertension, and multipleCWSs have been found to indicate rapidly advancing retinopathy.24 Theimportance of CWSs as a feature of severe nonproliferative retinopathy andtheir use as a predictor of proliferative lesions was questioned by the ETDRS.7 In our study a higher proportion of the less tightlycontrolled BP patients had CWSs at 4.5 and 7.5 years than the tightly controlledBP group, suggesting that these lesions were related to the level of BP.
There was no evidence of differences in rates of progression of thoseretinal lesions that we have analyzed between those allocated to captoprilor to atenolol therapy. Some have claimed that slowing of retinopathy progressionwould be optimized by the use of ACE inhibitors,25 butthese data suggest that it is reduction in BP per se that is the crucial intervention.
As expected, in type 2 DM, maculopathy was the most common reason forphotocoagulation. Although macular edema has sometimes been thought not tobe preventable, we, nevertheless, have demonstrated herein that there wasa 42% reduction in those in the tight control BP group. Concordant with thesedata on ETDRS step changes and individual features, we did not find any differencebetween the 2 policies of BP control, emphasizing that it is tight BP controlthat is more important than the pharmacological agent.
No patient became blind in both eyes. Blindness in 1 eye only occurredin 30 eyes—a small number that might be expected in a group of patientseven under close clinical monitoring. However, other visual loss is also important.Losing 3 lines in visual acuity is an important event and would mean thatone with initially normal vision would have difficulty with small print orfigures, and someone with even slightly reduced vision initially would havesignificant problems. The proportion of patients who lost this degree of visualacuity was significantly higher in the less tight BP control group comparedwith the tightly controlled BP group.
The question arises as to why high BP is so detrimental to the progressionof diabetic retinopathy. The retina has no functioning sympathetic nerve fibers,so that the control of blood flow is entirely by autoregulation. The normalautoregulatory response to high BP is vasoconstriction, tending to keep bloodflow constant. However, in patients with poorly controlled DM and retinopathy,blood flow is increased26 and this counteractsthe normal vasoconstrictive effect of raised BP. Furthermore, in long-standingDM, autoregulation is impaired.27 High BP,therefore, increases blood flow and, thus, by increasing shear stress willdamage vessel walls and will precipitate and worsen retinopathy. Rassam etal28 found that to normalize autoregulatoryfunction of the retinal vessels, both BP and blood glucose level had to becontrolled. This is also apparent in the observational studies within theUKPDS relating to microvascular disease generally.3 Wehave also demonstrated that tight control of BP is important for both primaryand secondary prevention of diabetic retinopathy.
High BP is detrimental to each aspect of diabetic retinopathy and atight BP control policy reduces the risk of clinical complications from diabeticeye disease.
Correspondence: David R. Matthews, FRCP,Oxford Centre for Diabetes, Endocrinology, and Metabolism, Churchill Hospital,Old Road, Oxford OX3 7LJ, England (email@example.com).
Submitted for Publication: July 1, 2003; finalrevision received July 22, 2004; accepted July 22, 2004.
Financial Disclosure: None.
Funding/Support: This study was supported inpart by major grants from the United Kingdom (UK) Medical Research Council,London, England; British Diabetic Association; the UK Department of Health,London; National Eye Institute, National Institute of Digestive, Diabetes,and Kidney Disease in the National Institutes of Health, Bethesda, Md; theBritish Heart Foundation, London; Novo-Nordisk A/S, Copenhagen, Denmark; Bayer(Schweiz) AG, Zurich, Switzerland; Bristol-Myers Squibb, New York, NY; HoechstAG, Kehl, Germany; Eli Lilly & Co, Indianapolis, Ind; Lipha and FarmitaliaCarlo Erba, Milan, Italy. Other funding companies and agencies, the supervisingcommittees and all participating staff are acknowledged in an earlier paper.13
Acknowledgment: We acknowledge the help ofHung Cheng, FRCS, and Carol Hill in the preparation of this manuscript. Thecooperation of the patients and many National Health System and non-NationalHealth System staff at the centers is much appreciated.
The Participating UK Centers are: Radcliffe Infirmary, Oxford, England;Royal Infirmary, Aberdeen, Scotland; University Hospital, Birmingham, England;St George’s Hospital, ; Hammersmith Hospital, Hammersmith, England;and Whittington Hospital, London, England; City Hospital and Royal VictoriaHospital, Belfast, Northern Ireland; North Staffordshire Royal Infirmary,Stoke-on-Trent, England; St Helier Hospital, Carshalton, England; Norfolkand Norwich Hospital, Norwich, England; Lister Hospital, Stevenage, England;Ipswich Hospital, Ipswich, England; Ninewells Hospital, Dundee, Scotland;Northampton General Hospital, Northampton, England; Torbay Hospital, Torbay,England; Peterborough General Hospital, Peterborough England; ScarboroughHospital, Scarborough, England; Derbyshire Royal Infirmary, Derby, England;Manchester Royal Infirmary, Manchester, England; Hope Hospital, Salford, England;Leicester General Hospital, Leicester, England; Royal Exeter and Devon Hospital,Exeter, England.