1201-P: Glycemic Response Phenotypes on Metformin Monotherapy in Real-World Diabetes Care

Metformin is the most prescribed initial treatment for type 2 diabetes in many parts of the world. Glycemic response to metformin has been described based on glycemic change - a measure that ignores potential differences in glycemic trajectories or paths between the start and end of metformin monotherapy. We examined HbA1c trajectories over two years in adults treated with metformin and developed prediction models for treatment response. We studied adults treated with metformin alone for >30 days after diabetes diagnosis in the US VA Health System from 20to 2016 who had a minimum of two HbA1c measurements from 90 days prior to two years after the first metformin prescription (N=140,413) . We included all HbA1c measurements (487,6total) during two years after metformin initiation, censoring at addition of a 2nd diabetes medication, metformin discontinuation, nursing home admission, death, or 90-day metformin non-adherence (medication possession ratio <80%) . We used latent class mixed models to identify HbA1c trajectory classes, then used random forests machine learning to predict trajectory class membership based on demographic, laboratory, vital sign, and comorbidity data at the time of diabetes diagnosis. We found three HbA1c trajectories: stably low (89.7% of sample, mean HbA1c decrease from 7.2% to 6.6%) , brisk response (7.1% of sample, mean HbA1c decrease from 11.4% to 7.0%) , and non-response (3.1% of sample, mean HbA1c increase from 8.9% to 10.8%) . Of those in the stably low and brisk response classes at 2 years, 91% maintained HbA1c at ∼7% on metformin alone for 5 years after drug initiation. Random forests yielded prediction models that accurately predicted brisk response (91% accuracy) but not non-response (59% accuracy) . CONCLUSION: Three glycemic response phenotypes were evident by 6 months after metformin monotherapy initiation in real world data and persisted in the vast majority out to 5 years. Early HbA1c surveillance can inform diabetes treatment optimization with durable impact.