Physicians' Academy for Cardiovascular Education

Absolute risk reduction prediction model helps personalising CV medicine for DM patients

Kaasenbrood L et al., Circ Cardiovasc Qual Outcomes. 2016

 
Development and Validation of a Model to Predict Absolute Vascular Risk Reduction by Moderate-Intensity Statin Therapy in Individual Patients With Type 2 Diabetes Mellitus

The Anglo Scandinavian Cardiac Outcomes Trial, Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial, and Collaborative Atorvastatin Diabetes Study

 
Kaasenbrood L, Poulter NR, Sever PS, et al.
Circ Cardiovasc Qual Outcomes. 2016;9:213-221
 

Background

Guidelines recommend statin therapy for most type 2 diabetes (T2DM) patients, based on the average relative risk reduction (RRR) of MACE seen in major clinical trials [1-3]. In clinical practice however, the absolute risk reduction (ARR) based on individual patient characteristics, rather than the RRR, is important for treatment decision making [4].
The usual approach to ‘personalise’ clinical trial results relies on subgroup analyses, which have not shown any significant heterogeneity regarding the relative effect of statins in T2DM patients [1]. Moreover, in subgroup analyses, only one characteristic is studied at a time, although treatment effects are determined by a combination of patient characteristics [5-7].
In this study, a multivariable prediction model for ARR of MACE by statin therapy for individual T2DM patients was developed and validated, in an effort to translate the average relative effect of statin therapy from trial data to the individual T2DM patient. The model was derived from data of 2725 patients enrolled in the ASCOT-LLA study (atorvastatin 10 mg versus placebo) [8] and validated  in data from the ALLHAT-LLT study (n=3878, pravastatin 40 mg versus usual care) [9], and the CARDS study (n=2838, atorvastatin 10 mg versus placebo) [10]. The model was based on 8 clinical predictors: age, sex, smoking, SBP, dyslipidaemia, history of CV events, fasting plasma glucose level, and treatment allocation (statin/placebo).
 

Main results

  • There was a wide distribution in predicted 10-year ARR by statin therapy. ARR was lower than 2% for 13% of patients, which translates to 10-year individualised number needed to treat [iNNT] over 50 for 13% of T2DM patients.
    ARR >4 was seen in about 30% of patients, translating to 10-year iNNT < 25.
    Median ARR was 3.2% (IQR: 2.5%–4.3%; 95%CI: –1.4% to 6.8%.
  • The addition of treatment interactions did not improve the model performance, suggesting that the wide distribution in ARR was a consequence of the underlying distribution in CV risk in these trials.
  • Model calibration was adequate in both external data sets, and discrimination was moderate: ALLHAT-LLT c-statistic: 0.64; 95% CI: 0.61–0.67, CARDS c-statistic: 0.68; 95% CI: 0.64–0.72.
  • Net benefit of applying a prediction model for selective treatment of T2DM patients with a statin is higher than a strategy in which all patients are treated, especially for 10-year numbers willing to treat of ≤50.

Conclusion

In 3 large cohorts of T2DM patients, there was a wide distribution in ARR of MACE by statin treatment due to the distribution of CV risk at baseline. The application of a model based on routinely available patient characteristics resulted in good estimations of ARRs of MACE by statin therapy for individual T2DM patients. This approach may be of additional value in personalising CV medicine in T2DM patients.
 

Editorial comment [11]

In their editorial, Arnold and Kosiborod, provide an additional perspective regarding the value of the Kaasenbrood et al prediction model: “Although traditionally, the concept of personalised medicine has referred to genetic personalisation, this also applies to traditional clinical and demographic factors, which may affect the balance of risks and benefits a patient is expected to derive from a treatment.” (…). “Multivariable models can sometimes effectively quantify the heterogeneity of treatment benefit. In this way, patients can be identified who are at most (and least) likely to benefit from a treatment.”
”While this is a noble goal, the number of risk models that are created far exceed those used in clinical practice.”(…) “Importantly, for a model to be useful, it should (1)  be accurate at estimating risk, (2) be able to separate patients into distinct risk categories, (3) be able to be used with available data at the time of decision making, and (4) exert a substantial impact on a treatment decision.”(…) “Kaasenbrood et al sought to challenge the one size fits all approach to statins and diabetes mellitus and attempt to identify patients at sufficiently low cardiovascular risk to recommend against statin therapy for primary prevention. The authors were limited by the variables they were able to include as potential predictors and therefore were not able to construct the most accurate model. However, this strategy was necessary to allow for external validation, a critical step in model development for which the authors should be commended.”
 
Find this article online at Pubmed
 

References

1. Kearney PM, Blackwell L, Collins R, et al. Efficacy of cholesterol-lowering therapy in 18,686 people with diabetes in 14 randomised trials of statins: a meta-analysis. Lancet.2008;371:117–125.
2. Perk J, De Backer G, Gohlke H, et al; European Association for Cardiovascular Prevention & Rehabilitation (EACPR); ESC Committee for Practice Guidelines (CPG).The Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). European Guidelines on cardiovascular disease prevention in clinical practice (version 2012). Eur Heart J. 2012;33:1635–1701.
3. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2013;63:3024–3025.
4. Sheridan SL, Viera AJ, Krantz MJ, et al; Cardiovascular Health Intervention Research and Translation Network Work Group on Global Coronary Heart Disease Risk. The effect of giving global coronary risk information to adults: a systematic review. Arch Intern Med. 2010;170:230–239.
5. Kent DM, Rothwell PM, Ioannidis JP, et al. Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal. Trials. 2010;11:85.
6. Burke JF, Hayward RA, Nelson JP, et al. Using internally developed risk models to assess heterogeneity in treatment effects in clinical trials. Circulation. 2014;7:163–169.
7. Kent DM, Hayward RA. Limitations of applying summary results of clinical trials to individual patients: the need for risk stratification. JAMA.2007;298:1209–1212.
8. Sever PS, Poulter NR, Dahlöf B, et al. Reduction in cardiovascular events with atorvastatin in 2,532 patients with type 2 diabetes: Anglo-Scandinavian Cardiac Outcomes Trial–lipid-lowering arm (ASCOT-LLA). Diabetes Care.2005;28:1151–1157.
9. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. Major outcomes in moderately hypercholesterolemic, hypertensive patients randomized to pravastatin vs usual care: The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT-LLT). JAMA. 2002;288:2998–3007.
10. Colhoun HM, Betteridge DJ, Durrington PN, et al; CARDS investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet. 2004;364:685–696.
11. Arnold SV, Kosiborod M. Predicting the Benefit of Statins in Patients With Diabetes Mellitus: A Case of Perfect Being the Enemy of Good? Circ Cardiovasc Qual Outcomes. 2016;9:191-3.