HRT Timing x Body Maturation

The useful question is not “is it too late?” but “which tissue clock is still running, and how much does another year actually cost?”

This page is a compressed portal into the research folder under research/hrt-timing-body-changes/. The deep pass removed several popular but weak claims: no more “clavicle fuses at 25,” no “sternum finishes at 25,” no hard age-coded Tanner ceiling, and no unsupported universal percentage for testicular shrinkage. What remains are the claims the current primary literature can actually carry.

Three classes matter. Bone and voice are developmental-architecture systems with steep early cliffs. Skin, body composition, breast tissue, and gonadal function are partially salvageable adult endocrine systems. Hair sits in the cumulative-exposure class: there is no one closure date, but more years still cost something.
Counterfactual delay is system-specific. One year at age 13 can meaningfully change voice, beard recruitment, and facial dimorphism. One year at 31 usually does not change bone architecture much, but it can still worsen scalp loss, body-hair burden, fertility logistics, and photoaging.
01

Age x System Matrix

The full age-banded narrative lives in age-windows-matrix.md. This table is the quick reference. Read the cells as “what a delay usually costs at this age,” not as a claim that every person in the band is at the same Tanner stage.

Age band Still meaningfully open Closing Mostly closed / later surgery-dominant Main delay cost Key anchors
12-14 Voice prevention, beard/body-hair prevention, major skeletal optionality, pubertal muscle trajectory None uniformly closed Usually few High for voice, face, shoulders, hair Puberty biology; voice change studies; blocker/BMD cohorts
14-17 Remaining skeletal optionality, hair prevention, body composition Voice, long-bone growth, facial dimorphism Some voices already changed High for voice, beard, muscle accrual Harries; Sorensen; Klink/Vlot
17-20 Body composition, skin, hair preservation, fertility planning Late clavicle tail, some late thoracic maturation Voice, height, most hand/foot growth Moderate-high for hair, muscle, fertility PMID 26188638; PMID 38960911
20-23 Fat redistribution, lean-mass loss, skin, breasts, gonadal suppression Late clavicle tail only Voice and major bone structure Mostly cumulative hair/body-comp cost PMID 27572683; PMID 34415999
23-26 Adult endocrine response remains strong Scalp preservation advantage narrows Bone and voice Hair, skin, fertility Adult trans cohorts
26-30 Fat, muscle, skin, breast growth Mainly cumulative windows Structural face / voice / shoulder goals AGA, body-hair workload, skin history Klaver; de Blok
30-35 Body composition and skin still very live Scalp preservation, fertility logistics Bone and voice AGA and cumulative exposure O'Connell review; fertility series
35-40 Endocrine effects, fat, skin, some breast growth Hair preservation, modest breast ceiling Structural goals Hair, skin, bone-risk exposure Brincat; Wiik; bone follow-up data
40+ Body composition, skin, bone maintenance, sexual-function effects Fertility and scalp recovery Voice and bone architecture Cumulative hair/fertility/bone-health cost Late-adult endocrine and bone-risk framing
02

Bone and Skeleton

The decisive fusion signal is estradiol in both sexes. Morishima’s aromatase-deficient man and Smith’s estrogen-resistance case are still the classic human demonstrations that estrogen signaling, not testosterone by itself, closes plates (PMID 8530621; PMID 8090165).

Late marker, not magic loophole

Medial clavicle is one of the last routinely assessed epiphyses. In Ekizoglu’s thin-slice CT series, substage 3c first appeared at age 19 in both sexes (PMID 26188638). In Freiburg’s male CT cohort, only complete symmetrical stage 4 on both sides was a reliable marker of age >21 (PMID 38960911).

That supports a real late-teen / early-20s skeletal tail. It does not support “clavicle fuses at 25,” and it does not mean adult estradiol can noticeably narrow the bony shoulders.

What is actually closed by adulthood

Most hand, wrist, and long-bone growth is adolescent. Adult estradiol does not reopen fused plates, change adult height, or materially shrink hands and feet. Bone-density maintenance is a separate question from bone shape.

The sternum / ribcage claim was tightened on purpose: developmental sternebral fusion, manubriosternal fusion, and lifelong costal-cartilage mineralization are different processes. “Sternum ~25” was not defensible.

03

Voice and Larynx

This is the strongest early cliff in the folder. Testosterone puberty changes the instrument itself: laryngeal framework, vocal-fold length, tissue mass, and tract resonance.

Prepubertal overlap, then divergence

Sorensen and Horii measured children aged 7-15 and found median F0 values of 244 Hz in girls and 250 Hz in boys before major divergence (PMID 7558642). Adult reference studies center speech around about 190-225 Hz in women and about 105-135 Hz in men (PMID 7300286; PMID 3049278).

Why estradiol does not reverse it

The Endocrine Society guideline remains explicit: feminizing hormones do not raise pitch after testosterone puberty (PMID 28945902). The reason is structural. Later estradiol may affect hydration and comfort, but it does not shrink the laryngeal framework back down.

Therapy and surgery still work. A recent meta-analysis found benefit for both, and a modern Wendler series reported about +33 Hz speaking F0 and +50 Hz sustained-vowel F0 at 1 year (PMID 39963873; PMID 35634734).

04

Skin

Skin is hormone responsive in adulthood, but not on a blank canvas. Sebum changes fast; collagen and dermal thickness change more slowly; photodamage is partly independent of current hormone state.

Fast endocrine signal

Pochi’s classic work supports puberty-related androgen increases in skin-surface lipids and sebum biology (PMID 143498). In trans women, Giltay and Gooren found objective sebum reduction within the first treatment year.

Slower matrix signal

Brincat’s postmenopausal studies remain the best human analogue for estrogen effects on dermis. Treated women had mean skin collagen content about 48% higher than untreated controls in the 1983 study, and later work showed prevention or partial restoration of menopausal collagen loss (PMID 6416400; PMID 3828252).

That supports partial recovery. It does not support the idea that late estradiol erases decades of UV damage or acne scarring.

05

Hair

Hair timing runs in two directions: scalp follicles miniaturize under androgen exposure, while beard and many body fields terminalize. The same hormones do opposite things in different follicles because the follicles are biologically different.

Scalp

AGA is cumulative, not a one-time puberty event. Earlier suppression preserves more follicles; later treatment more often stabilizes than fully regrows. That is why a delay at age 28 still matters even if it no longer matters for bone.

Body hair

The strongest later-adult recruitment signal is not “every field keeps thickening equally forever.” It is that ear, nose, back, shoulder, and trunk fields often keep recruiting later than beard puberty. The 2023 GAHT hair review supports cautious wording: feminizing therapy may reduce body and facial hair and may improve AGA, but evidence is heterogeneous (PMID 37311161).

06

Body Composition and Bone Density

This is the strongest adult “not too late” system in the folder. Adult GAHT changes fat, lean mass, and visceral-to-subcutaneous patterning far more than many people assume.

Quantified adult changes

Klaver’s meta-analysis found average 12-month changes of about +3.0 kg body fat and -2.4 kg lean body mass in trans women (PMID 27572683). In a large prospective cohort, trans women gained 4.0 kg total body fat while mean visceral fat stayed near flat, lowering the VAT/TBF ratio by 17% (PMID 34415999).

Strength and bone

Wiik et al. found significant loss of muscle size and strength over 12 months, even though many measures remained above cis-female references after one year (PMID 31794605). For bone, the main pediatric caution is blockers without sufficient follow-on sex steroids: Dutch follow-up shows BMD Z-scores fall during GnRHa and recover after later GAHT, with transfeminine lumbar spine as the main residual concern (PMID 25427144; PMID 27845262).

07

Face: Soft Tissue vs Bone

The face is mixed. HRT can still move the soft-tissue layer enough to matter a lot. It does not reliably move the bony layer enough to replace structural surgery where bone is the issue.

What HRT can move

Skin oiliness, facial-fat compartments, cheek softness, and masseter heaviness can all shift. That is why many adults see meaningful facial feminization over the first 6-24 months even though the skeleton is unchanged.

What remains structural

Brow bossing, deep nasion, gonial width, chin breadth, and thyroid-cartilage prominence are mostly bone or cartilage. Ousterhout and Capitán remain the classic references for why forehead surgery exists: these are structural targets, not skin targets.

08

Genital, Breast, and Reproductive

This system is partially reversible but not all-or-none. Hormones commonly suppress fertility and gonadal activity. They do not guarantee immediate permanent sterility, and they do not guarantee later recovery either.

Fertility and testicular histology

In Jiang’s orchiectomy series after more than a year of hormones, germ cells were still present in 81% of testes and spermatids in 40% (PMID 31310772). That is severe suppression, not universal complete absence.

Recovery after stopping GAHT is possible in some patients, but timelines are heterogeneous and one recovery series included a patient who required testicular sperm extraction after 17 months. Banking first remains the honest recommendation.

Breast growth and vaginoplasty planning

de Blok’s 3-year cohort showed continued breast-volume increase beyond year 1. Average gain was about 72 mL per breast, reaching roughly 100-101 mL by year 3, while breast-chest difference plateaued earlier (PMID 33206172).

Early suppression can reduce tissue available for classic penile inversion, but peritoneal-flap series show that genital hypoplasia does not remove vaginoplasty from the option set (PMID 36729740).

09

Graphs and Data

The plotting manifest now lives in graphs-and-data.md. The most useful curves are not decorative; they correspond to the real tissue classes in the folder.

Best graph candidates

  • F0 across life: childhood overlap around 244-250 Hz, then the sharp male pubertal drop toward ~120 Hz.
  • Medial clavicle maturation: stage 3c first at 19; reliable symmetrical stage 4 as a >21 marker in thin-slice male CT.
  • Adult GAHT body comp: +3.0 to +4.0 kg fat mass, -2.4 kg lean mass, -17% VAT/TBF ratio over the first year.
  • BMD under blockers then GAHT: Z-score dip during GnRHa, then recovery after sex-steroid replacement.
  • Breast volume over 3 years: early anthropometric plateau with slower continued volume increase.

Figure caveats

  • Many curves are assembled from different cohorts, not one longitudinal dataset.
  • Voice norms depend on speech task and language.
  • Body-hair and skin curves are less standardized than scalp-AGA or DXA data.
  • Forensic staging data are not direct treatment-effect data.
10

Myth Checks

Claims removed or tightened

  • “Clavicle fuses at 25.” Too tidy. The useful data are stage-based, not slogan-based.
  • “Sternum finishes around 25.” Removed. It collapses distinct developmental and degenerative processes.
  • “Estradiol shrinks shoulders.” False for the bony frame. Soft tissue and posture can change.
  • “Ribcage keeps widening a lot under T after 18.” Not well supported beyond the late-adolescent / early-adult tail.
  • “Older starters usually cap at Tanner 3.” Not established by strong longitudinal primary evidence.

The honest summary

Voice and hard skeleton are the steepest early cliffs. Body composition, skin, and breast tissue remain more salvageable later. Hair is the clearest cumulative-exposure system: there is no single closure date, but years still matter.