Moving Toward the Implementation of Precision Medicine Needs Highly Discriminatory, Validated, Inexpensive, and Easy-to-Use Prediction Models

Development and Validation of the Life Expectancy Estimator for Older Adults with Diabetes (LEAD): the Diabetes and Aging Study

BACKGROUND

Estimated life expectancy for older patients with diabetes informs decisions about treatment goals, cancer screening, long-term and advanced care, and inclusion in clinical trials. Easily implementable, evidence-based, diabetes-specific approaches for identifying patients with limited life expectancy are needed.

OBJECTIVE

Develop and validate an electronic health record (EHR)-based tool to identify older adults with diabetes who have limited life expectancy.

DESIGN

Predictive modeling based on survival analysis using Cox-Gompertz models in a retrospective cohort.

PARTICIPANTS

Adults with diabetes aged ≥ 65 years from Kaiser Permanente Northern California: a 2015 cohort (N = 121,396) with follow-up through 12/31/2019, randomly split into training (N = 97,085) and test (N = 24,311) sets. Validation was conducted in the test set and two temporally distinct cohorts: a 2010 cohort (n = 89,563; 10-year follow-up through 2019) and a 2019 cohort (n = 152,357; 2-year follow-up through 2020).

MAIN MEASURES

Demographics, diagnoses, utilization and procedures, medications, behaviors and vital signs; mortality.

KEY RESULTS

In the training set (mean age 75 years; 49% women; 48% racial and ethnic minorities), 23% died during 5 years follow-up. A mortality prediction model was developed using 94 candidate variables, distilled into a life expectancy model with 11 input variables, and transformed into a risk-scoring tool, the Life Expectancy Estimator for Older Adults with Diabetes (LEAD). LEAD discriminated well in the test set (C-statistic = 0.78), 2010 cohort (C-statistic = 0.74), and 2019 cohort (C-statistic = 0.81); comparisons of observed and predicted survival curves indicated good calibration.

CONCLUSIONS

LEAD estimates life expectancy in older adults with diabetes based on only 11 patient characteristics widely available in most EHRs and claims data. LEAD is simple and has potential application for shared decision-making, clinical trial inclusion, and resource allocation.

Circulating metabolites improve the prediction of renal impairment in patients with type 2 diabetes

INTRODUCTION

Low glomerular filtration rate (GFR) is a leading cause of reduced lifespan in type 2 diabetes. Unravelling biomarkers capable to identify high-risk patients can help tackle this burden. We investigated the association between 188 serum metabolites and kidney function in type 2 diabetes and then whether the associated metabolites improve two established clinical models for predicting GFR decline in these patients.

RESEARCH DESIGN AND METHODS

Two cohorts comprising 849 individuals with type 2 diabetes (discovery and validation samples) and a follow-up study of 575 patients with estimated GFR (eGFR) decline were analyzed.

RESULTS

Ten metabolites were independently associated with low eGFR in the discovery sample, with nine of them being confirmed also in the validation sample (ORs range 1.3-2.4 per 1SD, p values range 1.9×10-2-2.5×10-9). Of these, five metabolites were also associated with eGFR decline (ie, tiglylcarnitine, decadienylcarnitine, total dimethylarginine, decenoylcarnitine and kynurenine) (β range -0.11 to -0.19, p values range 4.8×10-2 to 3.0×10-3). Indeed, tiglylcarnitine and kynurenine, which captured all the information of the other three markers, improved discrimination and reclassification (all p<0.01) of two clinical prediction models of GFR decline in people with diabetes.

CONCLUSIONS

Further studies are needed to validate our findings in larger cohorts of different clinical, environmental and genetic background.

A Serum Resistin and Multicytokine Inflammatory Pathway Is Linked With and Helps Predict All-cause Death in Diabetes

CONTEXT

Type 2 diabetes (T2D) shows a high mortality rate, partly mediated by atherosclerotic plaque instability. Discovering novel biomarkers may help identify high-risk patients who would benefit from more aggressive and specific managements. We recently described a serum resistin and multicytokine inflammatory pathway (REMAP), including resistin, interleukin (IL)-1β, IL-6, IL-8, and TNF-α, that is associated with cardiovascular disease.

OBJECTIVE

We investigated whether REMAP is associated with and improves the prediction of mortality in T2D.

METHODS

A REMAP score was investigated in 3 cohorts comprising 1528 patients with T2D (409 incident deaths) and in 59 patients who underwent carotid endarterectomy (CEA; 24 deaths). Plaques were classified as unstable/stable according to the modified American Heart Association atherosclerosis classification.

RESULTS

REMAP was associated with all-cause mortality in each cohort and in all 1528 individuals (fully adjusted hazard ratio [HR] for 1 SD increase = 1.34, P < .001). In CEA patients, REMAP was associated with mortality (HR = 1.64, P = .04) and a modest change was observed when plaque stability was taken into account (HR = 1.58; P = .07). REMAP improved discrimination and reclassification measures of both Estimation of Mortality Risk in Type 2 Diabetic Patients and Risk Equations for Complications of Type 2 Diabetes, well-established prediction models of mortality in T2D (P < .05-< .001).

CONCLUSION

REMAP is independently associated with and improves predict all-cause mortality in T2D; it can therefore be used to identify high-risk individuals to be targeted with more aggressive management. Whether REMAP can also identify patients who are more responsive to IL-6 and IL-1β monoclonal antibodies that reduce cardiovascular burden and total mortality is an intriguing possibility to be tested.

The SMART Framework: Selection of Machine learning Algorithms with ReplicaTions - a Case Study on the Microvascular Complications of Diabetes

Over 34 million people in the US have diabetes, a major cause of blindness, renal failure, and amputations. Machine learning (ML) models can predict high-risk patients to help prevent adverse outcomes. Selecting the 'best' prediction model for a given disease, population, and clinical application is challenging due to the hundreds of health-related ML models in the literature and the increasing availability of ML methodologies. To support this decision process, we developed the Selection of Machine-learning Algorithms with ReplicaTions (SMART) Framework that integrates building and selecting ML models with decision theory. We build ML models and estimate performance for multiple plausible future populations with a replicated nested cross-validation technique. We rank ML models by simulating decision-maker priorities, using a range of accuracy measures (e.g., AUC) and robustness metrics from decision theory (e.g., minimax Regret). We present the SMART Framework through a case study on the microvascular complications of diabetes using data from the ACCORD clinical trial. We compare selections made by risk-averse, -neutral, and -seeking decision-makers, finding agreement in 80% of the risk-averse and risk-neutral selections, with the risk-averse selections showing consistency for a given complication. We also found that the models that best predicted outcomes in the validation set were those with low performance variance on the testing set, indicating a risk-averse approach in model selection is ideal when there is a potential for high population feature variability. The SMART Framework is a powerful, interactive tool that incorporates various ML algorithms and stakeholder preferences, generalizable to new data and technological advancements.