Combination T4/T3: A Normal TSH That Doesn't Feel Normal
A choir teacher's thyroid labs have been normal for a year, and she still can't keep pace with her own rehearsals. The disagreement isn't whether her labs are adequate — it's whether a normal number settles the question of whether her hormone replacement actually is.
Renata S., a 52-year-old high school choir teacher, has run the same three-choir spring concert cycle for eleven years without needing to sit down mid-rehearsal — until this year, when she started needing to lean on the piano twice a class period and losing her count partway through a phrase she's conducted a hundred times before. She was diagnosed with Hashimoto thyroiditis six years ago and has been on a stable levothyroxine dose since, with a TSH that has held steady across three checks this past year: 1.8, 2.1, and 1.6 μIU/mL, all comfortably inside the normal range, alongside a free T4 of 1.3 ng/dL sitting mid-range as well. Her fatigue and mental fog have been present for eight months now, enough that she has already worked through the standard alternative explanations with her primary care doctor: a normal complete blood count and ferritin rule out anemia, a negative sleep study rules out obstructive sleep apnea, and a PHQ-9 that doesn't clear the threshold for depression. She found an online thyroid support community over the winter and has come in today asking specifically about combination T4/T3 therapy, a term she can pronounce more confidently than her doctor expected.
The honest complication is that her labs are the exact picture combination-therapy trials were built to test, and those trials mostly came back unhelpful. A meta-analysis pooling eleven randomized trials of combination therapy against levothyroxine alone found no consistent advantage in quality of life, mood, or cognitive testing across the pooled population — Renata's normalized-TSH, still-symptomatic picture is not an edge case those trials missed, it's close to the exact population most of them enrolled. What the pooled result can't answer is whether SHE individually sits in the minority who do respond: a genotype at the DIO2 gene, which governs how efficiently peripheral tissue converts T4 into active T3, was linked by Panicker and colleagues, in a randomized crossover trial, to a real, blinded preference for combination therapy among carriers of one common variant — a biologically specific reason some patients on adequate-by-TSH monotherapy may still be under-converting locally, not just reporting a placebo response to a popular idea.
Reviewing a normal TSH that isn't solving the problem
I'd offer her a time-limited trial of combination therapy. The current ATA guidance, updated by Jonklaas and Bianco, doesn't recommend it as a default, but it does allow exactly this trial in a persistently symptomatic patient who has already had alternative causes ruled out — which she has. Her case is close to the profile the DIO2-polymorphism research describes: a patient whose TSH looks adequate on paper but whose peripheral conversion may not be keeping up.
I'd be cautious. The Grozinsky-Glasberg meta-analysis pooled eleven randomized trials of exactly this comparison and found no consistent benefit on mood, quality of life, or cognition — in plain terms, the best available evidence, taken as a whole, doesn't support combination therapy working better than monotherapy for the average patient in her position.
You're leaning on the DIO2 polymorphism data as though it applies to her specifically, but Panicker's finding came from a single crossover trial identifying a subgroup by genotype — we don't know her genotype, and offering the therapy on the strength of a subgroup finding she hasn't been shown to belong to risks treating a hypothesis as a diagnosis.
Both of you are right about different parts of this. The pooled trial evidence really is unimpressive, and the genotype data really is subgroup-level, not a personalized prescription. But a properly structured trial doesn't require either of those to be settled first: divided, lower-dose liothyronine added to her existing levothyroxine, monitored for a supraphysiologic T3 trough rather than dosed once daily the way early combination trials often did, with a defined eight-week endpoint and a specific symptom measure agreed in advance. If it doesn't move that measure, we stop and don't relitigate it every visit going forward.
Agreed: an eight-week trial of low-dose divided liothyronine added to her stable levothyroxine, with a symptom measure specified in advance and a trough T3 level checked to confirm dosing stays in physiologic range.
Not agreed: what happens if she reports feeling better but the symptom measure doesn't move. The endocrinologist would extend the trial, trusting her reported experience over an imperfect instrument; the primary care physician would stop on schedule regardless, worried about indefinitely continuing a therapy on the strength of an unmeasured impression rather than the agreed endpoint.