From 434 to 9.
From 33 to 42. From panic to peak.

The arc, in one paragraph.

In late 2023 I was in the best shape of my life — Apple Watch was estimating my VO₂max around 55, I was lifting heavy and recovering fast. Then came a Berkeley exchange semester, extreme stress, dizziness attacks so severe I ended up in the ER getting checked for a heart attack. Nothing structural. Back in Germany, my GP sent me to a therapist. The diagnosis in Q1 2025 was panic disorder. My nervous system was wrecked. I couldn't train. I went from 98 kg at the end of 2024 to 86 kg by the end of 2025 — most of it muscle, which for my height is not a good place to be.

At the same time, my gut was on fire. A colonoscopy showed first signs of Morbus Crohn. The February 2025 YEARS annual captured the bottom: VO₂max dropped below 33, calprotectin at 434 µg/g (reference ≤80), HRV percentile ranks near the floor across every axis. That's where the ledger starts. Everything below is what's happened since.

2024 was survive and heal again.
2025 was rebuild the foundation.
2026 is reach peak and stay there.

This drop is the first full-year view. I'm writing it partly because "open-sourcing a single person's recovery" sounds ridiculous right until you realise how much of this information lives only in expensive clinic reports that never see daylight. So here it is — imperfect, incomplete, but open. If a number, a protocol, or a failure here is useful to someone else, great. If it's motivating, even better.

Gut inflammation: 434 → 9 µg/g.

The single most dramatic number in the dataset, and the one I most want to be careful about. In February 2025, a stool test through Bioscientia Ingelheim returned a calprotectin value of 434 µg/g. The reference ceiling is ≤80. Calprotectin is a neutrophil-derived protein — the higher it goes in stool, the more active the intestinal inflammation (Pathirana et al., 2018). At that level, with the colonoscopy findings, the working diagnosis was early-stage Crohn's.

The protocol (to be clear: not a recommendation)

This is what I did. I'm listing it because transparency is the point, not because I'm claiming it will work for anyone else. Severe gut inflammation warrants a gastroenterologist, not a blog post.

• Anti-inflammatory diet — eliminated refined sugar, seed oils, gluten, dairy; prioritised wild-caught fish, olive oil, steamed vegetables, fermented foods at tolerance. Strong evidence base in IBD.
• Six weeks strict ketogenic — therapeutic keto for gut inflammation has thin but non-zero evidence (Paoli et al.). I used it as a reset, not a permanent pattern.
• Zinc-L-carnosine (75 mg BID) — actual RCT support for mucosal healing (Mahmood et al., 2007).
• Curcumin / turmeric — mixed but real evidence in active UC (Goulart et al., BMJ Open 2019; see also Cochrane CD008424 2020 update).
• L-glutamine — mechanistic rationale, poor clinical evidence. I added it anyway. Honest note: this one is closer to folklore than medicine.

The stack as a whole worked. I can't tell you which element did the heavy lifting, and neither can anyone else — that's a confound I'm explicit about. What I can tell you is: calprotectin went from 434 to 9, symptoms resolved, I stayed off biologics, and I'm still monitoring every six months because a single good reading isn't a cure.

Nervous system: HRV, every axis, climbing.

The panic disorder was the invisible centre of the 2024–25 collapse. Training on an activated sympathetic nervous system is how you burn through muscle without gaining fitness — and that's exactly what I did for six months before I got the diagnosis. The February 2025 HRV profile from YEARS was near the floor of the reference population across all five measured axes. By January 2026 every axis had climbed. Percentile rank vs. reference group, Feb 2025 → Jan 2026:

Higher percentile = healthier vs age- and sex-matched reference group. All five rose Feb 2025 → Jan 2026. Most still below the 50th percentile — direction is right, absolute level has a long way to go.

Every axis improved. Most are still below the median (50th percentile is "average"), and Dynamik 1 at the 7th percentile is still a concern. The direction is right; the absolute level has a long way to go. Interventions that drove the change: therapy, removing the stressor (Berkeley ended), jogging returning mid-2025, strength training returning end-of-2025, conscious breath-work and meditation, and — honestly — time. Panic disorders don't resolve on a stopwatch.

"You can't out-lift a nervous system in alarm mode. The HRV numbers told me to stop training before the labs did."

Cardio & lipids: ApoB moving, Lp(a) locked.

The interesting split in this dataset: everything modifiable in the lipid panel moved in the right direction. The one thing that is not modifiable — Lp(a) — stayed where genetics put it.

The clinical summary from YEARS put it precisely: "Lebensstil wirkt messbar, reicht aber bei hohem Lp(a) nicht aus." Lifestyle works — measurably — but at this Lp(a) level it isn't enough. Lp(a) is ~80 % genetically determined (Kronenberg, 2020; ESC/EAS 2022), and mine is in the high-risk zone. That's why the next intervention isn't another supplement — it's Ezetimib 10 mg/d, adding to the ApoB-lowering pressure without disturbing metabolism. Goal: ApoB ≤70 mg/dL. The question isn't whether to start preventive therapy early given a fixed Lp(a); it's when, and the evidence on cumulative LDL-years is clear enough that "earlier" is the defensible answer (Ference et al., 2017).

Muscle & fitness: back on the field.

The best-feeling category of the year, and the most painful one to look at honestly. I lost 12 kg of mostly-muscle between late 2024 and late 2025. The VO₂max drop speaks for itself: Apple Watch had me around 55 before Berkeley; the first YEARS test in February 2025 returned >33 ml/kg/min — a collapse. By January 2026 I was at 42.0 ml/kg/min (98 % of predicted), which is good but not where I started.

Grip strength is one of the better single-number proxies for all-cause mortality and cognitive trajectory (Cui et al., 2021), which is why I take it seriously. Both hands up, right/left asymmetry closing — that's a rebuild signature, not a cosmetic one.

Metabolic: insulin sensitivity up, HbA1c drifting.

The awkward pair of numbers this year: HOMA-IR improved from 2.8 to 2.3, fasting insulin dropped from 14.2 to 11.3 µIU/mL, OGTT remained clean — and yet HbA1c drifted from 5.5 % to 5.7 %. In isolation, 5.7 % is the pre-diabetes threshold. In context — OGTT normal, fasting glucose 84 mg/dL, insulin sensitivity improving — it's almost certainly a non-glycaemic artefact (RBC turnover, training-related), and the clinical interpretation from YEARS was "do not over-weight the HbA1c in isolation." I'll retest in H2 2026 and see.

The full H1 2026 panel.

Every marker I track half-yearly, with the previous reading in brackets where available. This is the complete lab payload for Jan 21, 2026, from YEARS × Bioscientia Berlin. Reference ranges in the tooltip or the full PDF (email me for it).

What didn't move — and why I'm saying so.

The point of an open ledger is not to parade the wins. Things that didn't move, or moved the wrong way:

• Lp(a) — flat, as expected. Lifestyle will never touch this meaningfully. Waiting on the next generation of Lp(a)-specific therapies, but in the meantime the response is to crush ApoB lower to compensate.
• Vitamin D — 42 → 41 ng/mL, essentially stable, despite supplementation. Target is 60 for bone density context. Response: dose up to 4000 IU/day plus 200 µg K2, retest in three months.
• HbA1c — drifted 5.5 → 5.7 % against a picture of improving insulin sensitivity. Watching, not acting.
• Homocysteine — 12.9 µmol/L, new finding this annual, flagged as elevated. B-complex (B6/B12/folate) starting. Retest in three months.
• HRV absolute levels — direction is right, absolute rank is still low on most axes. Nervous system recovery is measured in years, not quarters.

What's running through H2 2026.

The active intervention stack as of April 2026:

1. Ezetimib 10 mg/d — ApoB reduction. Target ≤70 mg/dL. First retest at the H2 2026 mid-year.
2. Vitamin B-complex (B6/B12/folate) — homocysteine target <10 µmol/L. Retest at 3 months.
3. Vitamin D3 4000 IU + K2 200 µg daily — 25-OH-D target 60 ng/mL.
4. Omega-3 EPA-dominant 2–3 g/d — Omega-3 Index target 10–12 %.
5. Saffron extract (Affron) 30 mg/d, 6 weeks — mood / stress axis, modest evidence (Marx et al., 2019).
6. Strength training — progressive, 4×/week, hitting PRs again.
7. HRV-informed cardio — morning reading, pull volume on low-HRV days. The one habit from the 2025 rebuild plan I am religious about.
8. Continued therapy + breath-work — because the HRV ranks still say the nervous system has further to go.

Next drop: H2 2026, covering the mid-year YEARS retest (scheduled July 2026), first Ezetimib retest, and the second calprotectin retest. If calprotectin stays <50 and ApoB crosses under 70, that's two of the bigger line items closed. If it doesn't — I'll tell you.

All labs: YEARS clinic (Berlin, Dr. Alexandru Ardelean) via Bioscientia · plus aware.app / MDI Limbach (baseline Jun 2024), Quest Diagnostics (Oct–Dec 2024 US), and Bioscientia Ingelheim (gut panel). Want the raw reports? niko@nikohems.de.