When retinoids stop causing peeling and redness, most users conclude the drug has stopped working. It hasn't. Your skin upregulates CYP26 enzymes that metabolize retinoic acid faster, cutting irritation without cutting clinical benefit. Long term studies show collagen synthesis and wrinkle improvement continuing well past the adaptation phase, including at year four.
Imagine you move to a city famous for its noise. The first week, traffic keeps you up at 3 a.m. By month three, you sleep through it. The noise hasn't changed. You have.
That's what happens when your tretinoin stops burning.
Most people read the absence of peeling as failure. The tube must have degraded. The prescription must have weakened. Something has grown accustomed to uselessness. So they increase the dose, switch to a stronger brand, or quit. All three moves misread the biology.
What Do We Mean by Tolerance?
Pharmacologists use the word "tolerance" to describe one thing: a drug producing less effect at the same dose over time. Beta blockers, benzodiazepines, and opioids genuinely lose potency as receptors downregulate. That isn't what happens with retinoids. What diminishes isn't the clinical benefit. What diminishes is the inflammatory response your skin mounts against a molecule it initially treats as a disruption signal.
These aren't the same thing.
Conflating them is how users end up abandoning an effective treatment at the exact moment it begins to work reliably. It's also why dermatologists see patients quit tretinoin at week ten, eleven, and twelve: precisely when the adaptation phase completes and the irritation resolves, and precisely when continuing would yield the most benefit.
Here's the underlying mechanism. Tretinoin binds to retinoic acid receptors. These come in three forms: RAR alpha, RAR beta, and RAR gamma. They function as transcription factors. Once activated, they travel into the cell nucleus and switch specific genes on or off. The genes they regulate govern three distinct processes: cell turnover in the epidermis, procollagen synthesis in the dermis, and suppression of the matrix metalloproteinases that degrade existing collagen, particularly the collagen degrading enzyme MMP 1 that UV exposure induces. None of that machinery slows when the peeling stops.
The redness, flaking, and sting belong to a separate pathway entirely. Your skin reads a high concentration of retinoic acid as a disruption signal and mounts a local inflammatory response. That response fades over weeks and months. The therapeutic response doesn't.
Week ten is usually when the adaptation completes. Not when the drug stops working. When it starts being tolerable.
The most shared tretinoin content on social media concentrates on weeks two through eight: the dramatic flaking, the purge, the peeling skin in ring light. That phase is visually interesting. When it ends, the content ends with it. Nobody posts at week twelve: no visible reaction, drug still working, skin feels normal. That's not compelling to watch. It doesn't photograph. But it's the most important part of the story, because it's precisely when most people who didn't quit are now on track to see the year two results, and it's when most people who did quit were on the exact same trajectory.
If you're uncertain whether your current retinol product is even delivering therapeutic concentrations, where retinol appears in your ingredient list tells you more than the front panel claim, and a product that doesn't deliver won't cause the early irritation that signals the adaptation phase has begun.
The Michigan Experiments
In the early 1990s, researchers at the University of Michigan Department of Dermatology were trying to answer a harder question than whether tretinoin works. They wanted biopsy evidence. They needed to see what was actually happening inside the dermis at the cellular level, not just what patients reported seeing in the mirror.
Their work, published in the New England Journal of Medicine in August 1993, was controlled and precise. Patients with photodamaged facial skin had matched sites treated with either tretinoin or vehicle. Biopsies were taken before and after. The finding: collagen I formation was 56% lower in photodamaged skin than in sun protected skin on the same patient. After tretinoin treatment, collagen I formation in the treated site rose by 80%, while the vehicle treated control site showed a 14% decrease over the same period.
That's a 94 percentage point spread.
What circulates less than the 80% figure is the follow on observation: the collagen improvements kept accumulating past the point when patients stopped reporting significant irritation. The adaptation phase and the therapeutic effect were running on entirely different timelines. Your skin wasn't getting used to doing less. It was getting used to the discomfort of doing more.
Subsequent work from the same Michigan group, notably John Voorhees and Gary Fisher publishing through the late 1990s, identified a key molecular node. UV exposure induces activator protein 1, a transcription factor abbreviated AP 1, which suppresses procollagen transcription. Tretinoin inhibits AP 1. This means tretinoin's mechanism isn't simply adding more collagen. It's closer to removing the signal that was stopping collagen production in the first place. That distinction matters for understanding why the benefit accumulates slowly and why stopping early is costly: you're not just pausing a process, you're reinstating the UV suppression signal the drug had been holding back.
What Happens at the Cellular Level?
Retinoic acid doesn't accumulate unchecked in skin. Your body regulates it through a family of metabolic enzymes: cytochrome P450 26, or CYP26. The family has three members: CYP26A1, CYP26B1, and CYP26C1. Their job is clearance. They convert retinoic acid into inactive metabolites and maintain its intracellular concentration within the homeostatic range, enough to run essential signaling without destabilizing gene expression.
When you apply tretinoin, you push retinoic acid concentration above that set point. CYP26 expression ramps up in response. A 2009 study in Biochimica et Biophysica Acta confirmed this directly in human epidermal keratinocytes: retinoic acid at 1 micromolar concentration induced CYP26A1 and CYP26B1 expression along with related binding proteins. The skin builds metabolic capacity proportional to the retinoic acid load it receives.
More enzyme means faster clearance.
Faster clearance means less retinoic acid available to activate the surface receptors driving the inflammatory cascade. The irritation subsides.
Here's the key distinction. CYP26A1 expression in skin is concentrated in basal keratinocytes, the cells of the epidermis. Dermal fibroblasts, the cells responsible for collagen synthesis, sit below that epidermal layer in the dermis proper. The clearance machinery that attenuates irritation in keratinocytes operates in a different cellular compartment than where the collagen building signal runs. The dose of retinoic acid reaching dermal fibroblasts stays sufficient to drive therapeutic benefit even as the epidermal irritation response winds down.
Two cell types. Two adaptive responses. Different outcomes.
"Some of my patients find retinoic acid too irritating, and experience skin flaking or stinging a few days after using it," wrote Dr. Whitney Bowe, MD, board certified dermatologist, on her skincare practice blog. Her clinical recommendation: start at a low concentration, buffer the skin with a moisturizer on early applications, and give the skin time to adapt. The adaptation isn't failure. It's the destination.
A 2022 review in Advances in Therapy frames this explicitly: the process represents the skin achieving a new homeostasis with the drug, not becoming resistant to it. Homeostasis, not resistance. That language shift changes how patients should think about what's happening to their skin.
If you're combining actives during the adaptation phase, the evidence on azelaic acid paired with tretinoin is more permissive than most forum consensus suggests, and understanding what the adaptation phase looks like helps distinguish normal retinoid response from a genuine incompatibility signal.
What the Long Term Data Shows
We rarely get four year randomized data on any topical cosmetic active. With tretinoin, we have it.
A long term efficacy and safety study, one of the few published at this duration, tracked patients over four years of tretinoin use. More than 80% showed progressive and then sustained improvement in photoaging features: fine and coarse wrinkles, mottled pigmentation, and sallowness. Perivascular inflammation, the tissue level marker of active irritation, dropped from 21% of patients at baseline to 8% at year four. The irritation resolved. The clinical improvement didn't.
The drug still worked.
A 2022 systematic review in the International Journal of Women's Dermatology consolidated the randomized controlled trial evidence: tretinoin produced clinically significant improvement in wrinkling, hyperpigmentation, and sallowness starting within one month and continuing through 24 months. And in a 2025 comparative analysis of topical interventions in Scientific Reports, when researchers ranked topical agents specifically for facial hyperpigmentation, tretinoin came out with the highest odds ratio of any tested agent: 4.78. That number includes patients who had already adapted, who had stopped peeling, and whose skin no longer showed the early phase inflammation that convinces most users the drug is working.
The drug leads the comparative evidence not despite tolerance developing, but while it has.
We track this pattern in Skinventry's Q1 2026 shelf scan data, specifically among users who logged retinoid progress notes alongside their product scans. The most common reason cited for stopping was that the retinoid had stopped doing anything. Those notes cluster in weeks 8 to 14, precisely when the CYP26 adaptation completes and the irritation phase ends. Our users quit at adaptation and call it failure.
Knowing what to expect in the early phase helps. The 12 week cellular arc of tretinoin walks through what's happening before adaptation sets in. For a comparison with retinol alternatives, what controlled studies actually found about bakuchiol is more qualified than most product reviews convey. And for anyone layering a retinoid under morning sunscreen, the actual SPF delivery from your daily product varies more than most retinoid protocols account for.
The Open Question
CYP26 upregulation explains why irritation fades. What it doesn't settle is the question I'd most like a randomized trial to answer.
After years of daily application, does CYP26 activity eventually rise high enough to reduce retinoic acid delivery below the therapeutic threshold at the dermis? The four year data says no, not in most patients. But the mechanism leaves the question structurally open. Some researchers have proposed CYP26 inhibitors as a path forward: compounds that block the clearance enzymes and preserve more retinoic acid per dose. The compound talarozole has been studied in exactly this context, and it appears in the same 2009 keratinocyte study that mapped CYP26 induction. Whether it offers meaningful benefit over standard dosing in patients with well adapted skin hasn't been answered in a controlled trial.
There's also the scheduling question. Many long term users shift to two or three nights a week after adaptation, usually from convenience rather than clinical guidance. Does a reduced schedule keep CYP26 expression lower, preserving a higher effective dose at the dermal level? Nobody has published that answer.
What I still don't know: whether the patients in the long term studies who showed sustained improvement at year four were on nightly dosing or had moved to maintenance frequency. That detail sits in protocol appendices I haven't been able to access. If you've found it, I'd like to hear.
The second question I'd want a trial to answer: at what point, if ever, does the skin fully adapt to a lower retinoid concentration and stop responding? The four year data says the response persists through continuous nightly use. But that's a specific population in a controlled setting. Most real users don't apply nightly every night for four years. They forget applications, they travel, they run out. Whether interrupted use changes CYP26 expression dynamics meaningfully, and whether that interruption is beneficial or harmful to long term outcomes, nobody has studied in a rigorous trial. It's a gap that matters more than most of the tolerance questions that do get studied.
Sources
- published in the New England Journal of Medicine in August 1993 · pubmed.ncbi.nlm.nih.gov
- A 2009 study in Biochimica et Biophysica Acta confirmed this directly in human epidermal keratinocytes · pubmed.ncbi.nlm.nih.gov
- concentrated in basal keratinocytes · pubmed.ncbi.nlm.nih.gov
- wrote Dr. Whitney Bowe, MD, board certified dermatologist, on her skincare practice blog · drwhitneybowebeauty.com
- A 2022 review in Advances in Therapy · pmc.ncbi.nlm.nih.gov
- International Journal of Women's Dermatology · pmc.ncbi.nlm.nih.gov
- Scientific Reports · nature.com