Most users feel nothing acutely. Metformin is not a felt-effect compound at therapeutic doses. Plasma metformin doesn't cross the BBB meaningfully, so there's no central pharmacological action. What users report is downstream / peripheral:

• GI side effects dominate the felt-experience profile — bloating, loose stools, occasional nausea, metallic taste, mild reflux. Mostly first 2-4 weeks; tapering up resolves it in ~60-70% of cases. Extended-release (Glucophage XR / Glumetza) cuts GI complaints by roughly half. Empty-stomach dosing dramatically increases GI symptoms.

• Some active users report subjective exercise impairment — heart rate harder to climb, perceived exertion higher at given workload, slower recovery between intervals, sluggish "second wind." This aligns directly with the Konopka/Walton mechanism data. Most pronounced for endurance + hypertrophy training; less noticed for short-duration / skill work.

• Mild appetite reduction (GLP-1-mediated, sub-clinical magnitude relative to semaglutide). Modest. Not a primary weight-loss drug at standard doses — average ~1-3 kg loss over 6-12 months.

• No mood, sleep, or direct cognitive felt effect. Some users report slight "mental sluggishness" — likely related to B12 depletion or GI-mediated fatigue rather than direct CNS.

• Lactate-related fatigue at high training intensity: Anecdotal reports of "feeling capped" at lactate threshold. Mechanistically plausible (metformin can shift lactate metabolism) but small magnitude in young healthy users.

• Honest variability: ~20-30% experience meaningful GI symptoms in first month. ~5-10% never tolerate it and discontinue. ~50-70% feel nothing felt-wise once stabilized.

• Tolerance: Glycemic effect persists indefinitely. No pharmacological tolerance to AMPK pathway activation in clinical practice. UKPDS 34 demonstrated sustained HbA1c effect across 10+ years. Some adaptive downregulation of AMPK signaling has been observed in vitro but is not clinically meaningful.

• Cycling: Not standard practice. Daily continuous use is the clinical norm for T2D. Some longevity-community users cycle (e.g., 5 days on 2 off, or 3 weeks on 1 week off) — usually framed as "preserving exercise adaptation" or "avoiding tolerance" — but there is zero RCT evidence base for cycling protocols. Attia has discussed cycling around heavy training blocks; this is purely mechanism-extrapolated.

• Reset: Not applicable in the receptor-recovery sense. Discontinue and metabolic effects taper over days-weeks; glycemic effect mostly gone within 1-2 weeks of stopping.

• Stack-rotation use: Some users alternate metformin with berberine on alternating weeks/months to capture overlapping AMPK activation without continuous metformin exposure. No outcome data.

Synergistic with

• Vitamin B12 (methylcobalamin 1000mcg/day): Mandatory for chronic users. Non-negotiable. Hydroxocobalamin injection (1000mcg IM monthly) is the gold standard for documented deficiency.
• Methylfolate + B6 (P5P): Offsets homocysteine elevation from B12/folate disruption. The user already takes P5P 100mg in V4 stack — that's the right idea.
• GLP-1 agonists (semaglutide, tirzepatide): Standard combination in T2D. Additive glycemic control. Different mechanisms (incretin-mimetic vs hepatic glucose output reduction). The metformin + GLP-1 combination is the modern dual-therapy gold standard.
• SGLT2 inhibitors (empagliflozin, dapagliflozin): Different glycemic mechanism (renal glucose excretion). Can stack; modern T2D combination therapy.
• Acarbose / pioglitazone: Different mechanisms; can stack.
• Rapamycin (theoretical longevity stack): Both inhibit mTOR via different routes (rapamycin direct mTORC1 binding; metformin upstream via AMPK). Some longevity practitioners stack low-dose metformin daily with weekly rapamycin 5mg. Zero human outcome data. Theoretically attractive, practically unvalidated.
• NAC, taurine, magnesium, omega-3: Neutral co-administration; all in Dylan's V4 stack, all safe.

Avoid stacking with

• Iodinated contrast agents (radiology): Hold 48 hours pre/post for contrast CT. Risk: contrast-induced AKI → metformin accumulation → lactic acidosis. Standard radiology protocol.
• Heavy alcohol use: Increases lactic acidosis risk via hepatic lactate buildup. Chronic heavy drinking is a relative contraindication; binge drinking on metformin is a hard "no."
• Sulfonylureas / insulin without monitoring: Hypoglycemia risk multiplicative.
• Berberine (redundant rather than dangerous): Both AMPK activators. Stacking is mostly redundant — pick one. Berberine is the OTC alternative when sourcing metformin is hard or for those wanting to avoid Rx pharmacology.
• High-intensity training programs (the athlete concern): Konopka/Walton mechanism shows blunted adaptation. Not a "dangerous" interaction but a counterproductive one for performance athletes. Dylan's MMA + lifting load is exactly this profile.

Neutral / safe co-administration

• All of the user's V4 stack: NAC, citicoline, magnesium glycinate, fish oil/DHA, PS, rhodiola, theanine, glycine, vitamin D3/K2, curcumin, beta-alanine, vitamin C, creatine — all clean, no interactions.
• Modafinil, bromantane, Semax/Adamax, Selank, ALCAR, taurine, astaxanthin, tryptophan — all clean.
• Standard SSRIs/SNRIs — no interaction.

Metformin's interaction profile is unusually clean because it is not metabolized by CYP enzymes — it's excreted unchanged in urine via OCT2 + MATE1/2 transporters. The clinically meaningful interactions are transporter-mediated, not CYP.

Transporter-mediated (the real interaction class):

• Cationic drugs sharing OCT2/MATE transport (cimetidine, dolutegravir, ranolazine, vandetanib, trimethoprim): Can raise metformin plasma exposure by 20-40%. Usually manageable with dose adjustment; flag for clinicians.
• OCT1 inhibitors (verapamil, certain antiretrovirals): Reduce hepatic uptake of metformin, potentially reducing efficacy.

Renal-function-mediated:

• NSAIDs (chronic high-dose): ACE inhibitors / ARBs / diuretics (especially loop): All can reduce eGFR. Combined use needs renal monitoring; if eGFR drifts below 45, metformin dose should be halved; below 30, discontinued.

Acidosis risk additive:

• Topiramate, acetazolamide (carbonic anhydrase inhibitors): Mildly additive metabolic acidosis risk.

Glucose-modulating:

• Glucocorticoids, atypical antipsychotics, beta-blockers, thiazides: All can raise glucose; metformin efficacy may appear blunted.
• Insulin secretagogues + insulin: Hypoglycemia risk additive (relevant only in diabetics).

Notably clean:

• Hormonal contraceptives: No interaction.
• Psychiatric meds (SSRIs, SNRIs, mood stabilizers): No interaction.
• Modafinil, amphetamines, methylphenidate: No interaction (different metabolism routes).
• Anabolic steroids: No direct interaction (some PED users add metformin specifically to manage AAS-induced insulin resistance; the community-data block here shows multiple stacks doing exactly this).

Alcohol: Acute moderate alcohol — minimal effect. Chronic heavy drinking — increases lactic acidosis risk via hepatic lactate accumulation. Binge drinking — acute risk. Dylan's zero-alcohol baseline makes this non-applicable.

Metformin's pharmacogenomics are transporter-driven, not CYP-driven.

• OCT1 (SLC22A1): The primary hepatic uptake transporter. Multiple loss-of-function variants (rs12208357 R61C, rs34059508 G401S, rs72552763 420del, rs34130495 G465R) reduce hepatic metformin accumulation. ~20-25% of European-ancestry individuals carry at least one reduced-function allele; ~9% carry two. Functional consequence: blunted glycemic response to standard doses, especially at lower doses. Probably explains a meaningful fraction of metformin "non-responders."

• OCT2 (SLC22A2): Renal uptake. Variants affect clearance. rs316019 808T variant lowers renal clearance ~30%; reduces dose tolerance.

• MATE1 (SLC47A1) and MATE2 (SLC47A2): Apical efflux into urine. Variants here affect renal excretion + lactic-acidosis risk in renal impairment.

• For users in this archetype: Once 23andMe results land (~June 2026), OCT1 status is informative if metformin is ever considered. Not a deal-breaker for the SKIP verdict, but moves the expected dose and efficacy if circumstances change.

• No CYP polymorphism relevance (not CYP-metabolized).

• TCF7L2 (rs7903146): Strongest GWAS hit for T2D risk; carriers may benefit more from metformin in pre-diabetes (per DPP genetic sub-analyses). Relevant only if Dylan ever flags pre-diabetic.

• ATM, SLC22A3, GLUT2 variants: Smaller pharmacogenomic signals; not clinically actionable.

Recommendation for the user's archetype: Not currently sourcing — verdict is SKIP. If reconsidered post-bloodwork or post-30: AgelessRx-style longevity telehealth is the cleanest path (legit Rx + clinical oversight + B12 monitoring guidance). Indian generic is the cheap-and-functional path if Rx friction is too high. Mexico OTC is geographically constrained.

• Baseline (before starting): HbA1c, fasting glucose, fasting insulin (calculate HOMA-IR = glucose×insulin/405), eGFR + creatinine, full liver panel (ALT, AST, alkaline phosphatase, bilirubin), vitamin B12, folate, homocysteine, vitamin D, magnesium, lipid panel (total cholesterol, LDL-C, HDL-C, triglycerides), lactate if available, TSH for thyroid baseline.

• 3 months on novel use: B12, folate, homocysteine, HbA1c, eGFR, GI tolerability check (subjective). Compare to baseline.

• Annually on chronic use: B12 + folate + homocysteine (mandatory — supplement methylcobalamin 1000mcg/day from start of chronic use, recheck B12 to confirm levels rising). HbA1c, lipids, eGFR, creatinine. Lactate if symptomatic or in setting of acute illness.

• Subjective during use: GI tolerability (rate weekly first month, monthly after), exercise capacity (perceived exertion + heart-rate response at standardized loads), mood/energy (rule out covert B12 effect), peripheral neuropathy symptoms (tingling, numbness — B12 deficiency presentation).

• Post-discontinuation: Re-check fasting glucose and HbA1c at 4-8 weeks off; verify metabolic effect tapers as expected.

• For Dylan post-23andMe + post-bloodwork (June 2026): If HbA1c >5.6, fasting insulin >7, or HOMA-IR >2 (none expected at 20yo, but track annually) — then revisit the decision matrix. If OCT1 genotype shows reduced function, factor into any future dose decision.