GLP-1s: interactions, bone, micronutrients and the "microdosing" fad

Interactions that get missed

• Sulfonylureas and insulin: on the tirzepatide label, hypoglycemia < 54 mg/dL was 10.3% with a sulfonylurea vs 2.1% without¹.
• Levothyroxine: with oral semaglutide, "levothyroxine exposure was increased 33%"². In the original pharmacokinetic study, in 45 healthy volunteers, the increase was in total T4 exposure, and the authors suggested considering monitoring of thyroid parameters¹¹. A case on subcutaneous semaglutide after thyroidectomy needed a 25% levothyroxine dose reduction for suppressed TSH³.
• Warfarin: in 1,021 patients, time in therapeutic range fell 2.1 points after starting a GLP-1⁵, although in healthy volunteers oral semaglutide did not appreciably change warfarin exposure¹²; the tirzepatide label asks for monitoring of narrow-therapeutic-index drugs¹.
• Oral contraceptives: one 5 mg dose of tirzepatide reduced ethinyl estradiol Cmax by 59%¹.

An important nuance: slower gastric emptying does not always reduce exposure to oral drugs. A 2026 review found increased exposure to thyroxine, atorvastatin and ramipril on some labels, and warns that "a reduction in exposure should not be assumed"⁴.

Muscle and bone

A systematic review of 35 trials (Ann Intern Med 2026) found that a median 28.3% of weight lost on incretins was in muscle-based indices such as fat-free mass, lean mass or muscle (IQR 15.9–39.9%); 65% of studies exceeded the ~25% benchmark. And more importantly: no study reported objective physical function⁶. In the SURMOUNT-1 body-composition substudy (160 participants with DXA), about 75% of the weight lost on tirzepatide was fat and 25% lean mass, a proportion similar to the placebo group¹³.

For bone, in a four-arm trial, liraglutide alone was associated with lower hip and spine bone mineral density than exercise alone; with liraglutide plus exercise, bone density did not change versus placebo⁷. The overall picture remains open: a 2026 meta-analysis of 60 studies found no effect on bone density or fractures in the most-adjusted models, but did find a reduction in lean mass, with low certainty of evidence¹⁴. Another 2026 systematic review of 8 randomized trials concluded that the evidence is too sparse and heterogeneous to determine the effect on bone; the only semaglutide trial showed lower lumbar spine bone mineral density¹⁵.

Micronutrients

A 2026 review pooled six studies with 480,825 adults. Vitamin D deficiency was the most common abnormality: 7.5% at 6 months and 13.6% at 12; ferritin was 26–30% lower than with SGLT2 inhibitors, and more than 60% consumed less calcium and iron than required⁸. The data are mostly observational and do not prove causation⁸. A 2025 joint advisory from four US nutrition and obesity societies listed nutritional deficiencies from reduced intake and loss of muscle and bone among the challenges of treatment, and prioritized preventing them with an appropriate diet and resistance training¹⁶.

"Microdosing"

Patient communities, telemedicine programs and clinician anecdotes popularized using 25–50% of the recommended starting dose or longer intervals, even "counting clicks" on pens. A 2026 commentary sums it up: "high-quality evidence is essentially absent"⁹. Another paper links it to "research-grade" peptides bought online and to medspas¹⁰.