Microvascular Disease and Risk of Cardiovascular Events Among Individuals With Type 2 Diabetes: a Population-Level Cohort Study
Brownrigg JRW, MRCS1, Hughes CO, MRCS2, Burleigh D, MSc3, Karthikesalingam A, PhD1, Patterson BO, PhD1, Holt PJ, PhD1, Thompson MM, MD1, de Lusignan S, MD3, Ray KK, MD,MPhil 4*, Hinchliffe RJ, MD1*
1Division of Cardiovascular and Cell Sciences, St George’s University of London, London, UK
2Division of Surgery and Interventional Science, University College London, London, UK
3Department of Healthcare Management and Policy, University of Surrey, Guildford, UK
4Department of Primary Care and Public Health, Imperial College London, London, UK
*KKR and RJH contributed equally to this study.
Word count:
4206
Corresponding author:
Jack RW Brownrigg
Division of Cardiovascular and Cell Sciences
St George’s University of London
Cranmer Terrace
London
SW17 OQT
ABSTRACT
Background
Diabetes confers a 2-fold excess risk of cardiovascular disease (CVD), yet predicting individual risk remains challenging. The effect of total microvascular disease burden on CVD risk among individuals with diabetes is unknown.
Methods
A population-based cohort of patients with type 2 diabetes from the UK Clinical Practice Research Datalink was studied (n=49027). We used multivariable Cox models to estimate hazard ratios forthe primary outcome (cardiovascular death, non-fatal myocardial infarction or non-fatal ischaemic stroke) associated with cumulative burden of retinopathy, nephropathy and peripheral neuropathy among individuals with no history of cardiovascular disease at baseline.
Findings
During a median follow-up of 5·5 years, 2822 (5·8%) individuals experienced a primary outcome. Significant associations were observed for the primary outcomeindividually forretinopathy, peripheral neuropathy, and nephropathy after adjustment for established risk factors. The hazard ratios (with 95% confidence intervals) were 1·39 (1·09-1·76), 1·40(1·19-1·66), and 1·35(1·15-1·58), respectively. For individuals with one, two or three microvascular disease states versus none, the multivariable-adjusted hazard ratios for the primary outcome were 1·32 (1·16-1·50), 1·62 (1·42-1·85) and 1.99 (1·70-2.34), respectively. Similar trends were observed for cardiovascular death, all cause mortality and for hospitalisation for heart failure. For the primary outcome, measures of risk discrimination showed significant improvement when microvascular disease burden was added to models. In the overall cohort, the net reclassification index for US and UK guideline risk strata were 3.6% (p<0.001) and 3.8% (p<0.001), respectively.
Interpretation
The cumulative burden of microvascular diseasesignificantly impacts the risk of future cardiovascular disease among individuals with type 2 diabetes. Given the prevalence of diabetes globally, further work to understand the mechanisms behind this association and strategies to mitigate this excess risk are warranted.
Funding
Circulation Foundation
RESEARCH IN CONTEXT
Evidence before this study
We searched Medline and EMBASE for studies published from Jan 1, 2000, to Nov 1, 2015, with the terms “microvascular disease”, “cardiovascular disease”, “type 2 diabetes”, and MeSH equivalents. The search period was selected to reflect contemporaneous data immediately before, and following the introduction of routine screening for microvascular disease in 2004 for the UK Quality and Outcomes Framework. We reviewed observational studies and clinical trials examining the association between microvascular disease and cardiovascular outcomes in ≥200 individuals. We identified 19 studies reporting positive associations between retinopathy or nephropathy and cardiovascular disease,coronary events,ischaemic stroke,and heart failure.More limited data also support a positive association between neuropathy (cardiac autonomic neuropathy or peripheral neuropathy) and cardiovascular disease outcomes. A single study in a Chinese cohort evaluated the impact of disease in two microvascular beds, with reported hazard ratios of 1.69 (95% CI 0.99-2.89) for retinopathy alone and 2.25 (95% CI 1.40-3.63) with concomitant microalbuminuria.Although some good quality studies were identified, all were limited in their scope by either small sample size with individual studies reporting on fewer than 630 events each, by the inclusion of individuals with pre-existing cardiovascular disease, selection bias, or lack of adjustment for conventional risk factors and for the presence of disease in multiple microvascular beds.
Added value of this study
Based on a detailed review of the literature, this study is the first to examine the effect of disease in multiple microvascular beds in a large population cohort, with approximately 260 000 person years of exposure and 2822 first cardiovascular events. Our data reveal several important findings. The presence of isolated retinopathy, peripheral neuropathy, or nephropathy, independent of conventional risk factors, confer at least a similar risk of cardiovascular events (cardiovascular death, non-fatal myocardial infarction or non-fatal ischaemic stroke) as uncontrolled established risk factors including blood pressure (≥140/90 mmHg), HbA1c (≥7.0%) and low-density cholesterol (≥2.5 mmol/L). Individuals with disease in multiple microvascular beds were, in a “dose dependent fashion”, at the greatest overall risk, including for other endpoints such as hospitalisation for heart failure, cardiovascular death and all-cause mortality.
Implications of the available evidence
These data suggest that a continued broad assessment program for microvascular complications of diabetes has prognostic value for routine clinical care globally as it further risk stratifies people at higher cardiovascular risk than might be perceived, as well as providing morbidity specific to individual microvascular disease states. The inclusion of microvascular disease variables in cardiovascular risk algorithms resulted in a net correct reclassification of 3.6% of our cohort into higher- or lower-risk strata based on incident events.which is comparable if not slightly better than blood based biomarkers, but less than improvement observed with coronary artery calcium scoring. If information on microvascular disease were incorporated presently then 9·3% of individuals previously considered as eligible for moderate intensity statins in US guidance (predicted risk <7.5%) would be considered as candidates for high intensity statin therapy (observed risk 8·6%). Similarly, microvascular disease would reclassify 9·0% of individuals in a higher risk group (predicted risk ≥7.5%), currently considered eligible for high intensity statins, to a group who could be offered moderate intensity therapy (observed risk 6·3%). In reference to UK NICE guidance, of those currently considered ineligible for statin therapy (predicted risk <10%), 8·9% would be reclassified into a higher risk group with an observed event rate of 11·6%. Of individuals currently offered statin therapy (predicted risk ≥10%), 12·3% would be reclassified into a lower risk group with an observed 10-year event rate of 8·1%.Based on the current known prevalence of risk, in absolute terms this would represent a change in statin prescriptions for 10·6% of individuals with type 2 diabetes in the UK and 9·1% of those in the US, with accurate reclassification in 59·5% and 65·7%, respectively.
As the assessment of microvascular disease should be part of routine clinical practice among those with diabetes, our findings offer a simple, convenient and cheap method for improving risk prediction as compared to more expensive blood based biomarkers or non-invasive imaging modalities for better targeting preventive therapies. It might be possible to mitigate against this excess risk, as we observed that among those with multiple microvascular disease states, event rates were substantially lower when HbA1c, BP and LDL-C were better controlled. High microvascular disease burden could be used as a criteria to enrich future clinical outcome trials,identifying a very high risk cohort of patients who might derive greater absolute benefit from more intensive risk factor control with conventional or novel therapies. Finally, our observations should enthuse further research including a better understanding of the impact of microvascular disease with different cardiovascular outcomes.
INTRODUCTION
Diabetes confers a 2-fold excess risk of cardiovascular disease1 and substantial premature mortality from cardiovascular causes.2However, individuals with diabetes are not automatically considered as a coronary heart disease (CHD) risk equivalent and many guidelines now recommend absolute risk assessment prior to considering lipid modification therapy.3 Predicting individual risk remains challenging and external validation of available risk algorithms in diabetic populations show moderate performance at best,4highlighting the need for cheap and routinely available measures that identify those with higher absolute risk over and above established factors considered in contemporary risk algorithms.
Various microvascular disease states have been reported to be associated with risk of vascular disease, including cardiac autonomic neuropathy (CAN),5,6 retinopathy,7,8 nephropathy,9,10 and peripheral neuropathy.11 Despite frequently co-existing, robust population data evaluatingthe effect of cumulative microvascular disease burden on cardiovascular risk in diabetes is absent. The aim of this study was to investigate whether microvascular disease states alone or in unison are independently associated with cardiovascular disease (CVD), and furthermore to compare any strength of association with conventional risk factorsused in currentrisk equations. To assess this relationship, we used routine healthcare data from a large population-based cohort of individuals with type 2 diabetes free from CVD at baseline, with approximately 259 686 person years of follow up and 2689 first cardiovascular events.
METHODS
Data sources and cohort
The Clinical Practice Research Datalink (CPRD) comprises data on individuals from over 600 practices in England, providing a representative UK primary care population.12 CPRD contains information on anthropometric measurements, clinical diagnoses, laboratory tests and prescription data, coded with the Read Clinical Coding system. Information on retinopathy, nephropathy and peripheral neuropathy has been routinely collected in UK primary care following the introduction of a pay for performance initiative, the Quality and Outcomes Framework,13 in April 2004, which is linked to the National Institute for Health and Care Excellence (NICE) guidance on standards of care for patients in the UK including appropriate frequency of screening and risk factor control for those with chronic diseases.14
Individual patient data were linked across three datasets: the CPRD for demographic characteristics and, Hospital Episode Statistics (HES) and the Office for National Statistics (ONS) for the outcomes of interest. The HES are the English National Health Service administrative dataset and contain information on every hospital admission including diagnostic data, recorded as International Classification of Diseases, 10th revision (ICD–10), and procedural data based on the Office of Population, Census, and Surveys, version 4 (OPCS–4) codes. The ONS provide individual mortality records including cause of death (ICD–10).
The study start date was 1 April 2008 to allow for 4 years of quality data on microvascular disease status among participants. The data extract provided by CPRD included data on 49027 individuals aged 18 years and over with type 2 diabetes and complete information on the presence or absence of three microvascular diseases: retinopathy, nephropathy and peripheral neuropathy. Individuals were screened for the presence of diabetes using established criteria,15 and classified in accordance with methods described previously.16Diabetes was defined by fasting plasma glucose ≥7.0 mmol per litre (126 mg per decilitre), random plasma glucose ≥11.1 mmol per litre (200 mg per deciliter) or the use of glucose lowering medications, based on recommendations from the American Diabetes Association.15,17 In brief, classification of T2DM was performed according to the following criteria: specific diagnostic code for T2DM (Read code C10F; ICD–10 codeE11) with no contradictory code; and patients with a diagnosis of diabetes at ≥35 years of age with no insulin prescription within 1 year of diagnosis. Validation study of electronic health records using this approach corrected miscoding of diabetes type in between 6–8% of cases.16We excluded individuals with a prior history of any cardiovascular disease.
Definition of baseline variables
Anthropometric measurements and numerical data, including systolic and diastolic blood pressure, glycosylated haemoglobin, and cholesterol values were derived by taking the mean of the three most recent values in the 12 months prior to the study start date. In cases where three values were unavailable, the mean of two values was calculated. Values recorded more than 12 months prior to the study start were not considered.Smoking status was stratified into groups of never smoked, previously smoked and currently smoking at entry into the study. Code lists used to define microvascular disease states were developed in accordance with published guidance,18, 19 and are provided in the webappendix 1–3.Nephropathy was defined as microalbuminuria (a moderate increase in albuminuria: 3-30 mg/mmol, 30-300 mg/g, 30-300 mg/24h, or reagent strip urinalysis),20and or eGFR <60ml/min per 1.73m2.
Outcome ascertainment
The follow-up period extended to the study end: either December 2014, the date of patient transfer from an included practice, or death. The primary outcome was the time to first major cardiovascular event(an a priori composite of cardiovascular death, non-fatal myocardial infarction or non-fatal ischaemic stroke). Ischaemic stroke events were defined by ICD-10 codes (I63) in accordance with published guidance.21 We combined ischaemic strokes with unclassified strokes (I64) because previous studies have shown that 87% of unclassified strokes were ischaemic.22Information about cause-specific mortality and date of death was obtained through the established record linkage with ONS. Fatal myocardial infarction (MI)and ischaemic stroke were defined by primary cause of death (ICD–10 codes I21–I22 and I64 respectively).Patients were censored on the date of first primary outcome event. The pre-specified secondary endpoints werecardiovascular death (fatal MI or fatal ischaemic stroke), hospitalisation for heart failure and all-cause mortality. Study approval was granted by the Independent Scientific Advisory Committee of the Medicines and Healthcare products Regulatory Agency.
Statistical analyses
We defined clinical characteristics and outcome databoth overall and according to risk groups (absence of microvascular disease at baseline, or stratified by the number of prevalent microvascular disease states).All reported p values are two-sided. Adjusted hazard ratios and corresponding 95% confidence intervals were estimated with Cox proportional-hazards models.Adjustment in all models was performed for age, gender, on treatment systolic and diastolic blood pressure, high- and low-density cholesterol, HbA1c, body-mass index,duration of diabetes, smoking history (defined by either ex-smoker or current smoker status),antiplatelet therapy, lipid-lowering therapy, use of angiotensin converting enzyme inhibitor/ angiotensin receptor blocker, any treatment for blood pressure, ethnicity and index of multiple deprivation.The group free of microvascular disease at baseline were used as the reference category. Missing data for ethnicity and index of multiple deprivation were imputed using multiple imputation by chained equations in the “mice” algorithm in R, and these imputed data were used in the primary analysis.
We assessed differences in predictive accuracy of a model including established risk factors from the Framingham risk function for a first primary outcome event (model A),23 and the same model incorporating microvascular disease variables (model B). Model discrimination was assessed with the use of the C-statistic.24To evaluate the overall improvement in risk stratification with the addition of microvascular disease to fully adjusted models, we calculated net reclassification improvement (NRI)statistic and the integrated-discrimination-improvement (IDI) statistic.25Discrimination indices are reported across risk strata defined in both the American College of Cardiology (ACC)/ American Heart Association (AHA) treatment guidelines (lower risk <7.5%, higher risk ≥7·5%3for 10-year CVD risk and the UK National Institute for Health and Care Excellence (NICE) guidelines which consider higher risk individuals as those with≥10% 10-year risk of CVD.26Statistical analyses were performed with the use of R software version 15·2.
Role of the funding source
The sponsors had no role in the original protocol design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit the report for publication. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
RESULTS
Patient Characteristics
We identified a cohort of 49 027 individuals with type 2 diabetes, of whom just less than half were women. Baseline characteristics of the study population, both overall and according to microvascular disease burden, are shown in Table 1. Individuals with microvascular disease were more likely to have an adverse cardiovascular risk profile with significantly greater levels of HbA1c, systolic blood pressure andsmoking history. Age and duration of diabetes significantly increased in a linear fashion with increasing burden of microvascular disease. Exceptions includeda trend for morefavourable low-density lipoprotein cholesterolwith increasing burden of microvascular disease, likely related to the greater use of lipid-lowering therapy.A comparison of the demographic characteristics of individuals with a single manifestation of microvascular disease versus those without is provided in the webappendix 4.
Primary and Secondary Outcome Measures
Event ratesfor the primary outcome per 1000 person years in those without microvascular disease were 5·00 compared with 8.22, 10.12 and 10.04 among individuals with isolated retinopathy, nephropathyand peripheral neuropathy, respectively. Each microvascular disease state studied was significantly associated with the primary outcome, and remained so following adjustment for established risk factors and after excluding individuals with multiple manifestations of microvascular disease (Table 5 webappendix).Single manifestations of microvascular disease appear to confer at least as much risk as the failure to control conventional risk factor goals in adjusted analyses (webappendix 6–8). Microalbuminuria in the absence of low eGFR (<60ml/min per 1.73m2) was independently associated with the primary outcome (webappendix 9). Further adjustment for the number of antihypertensive treatments resulted in no qualitative difference in the hazard ratios for the primary outcome.
Figure 1shows the linear relationship between increasing burden of microvascular disease and the primary outcome (Panel A), cardiovascularmortality (Panel B), and hospitalisation for heart failure (Panel C), P for linear trend <0·001 for all. Analyses for all-cause mortality were qualitatively similar (webappendix 10); we found a 4·7-fold excess risk of death from any cause among individuals with three manifestations of microvascular disease compared with none (webappendix 11). Unadjusted event rates for the primary outcome among individuals free of microvascular disease at baseline and among those with one, two, or three microvascular disease states were 5·0, 9·8, 15·7 and 22·1 per 1000 person years, respectively.After adjustment for potential confounders, the hazard ratiosforthe primary outcome, cardiovascular death and hospitalisation for heart failure remained significant butwereattenuatedacross all three groups, suggesting that conventional risk factors account, in part, for the excess risk observed with cumulative burden of microvascular disease (Table 2).