5-FU vs Imiquimod vs Cryotherapy: Choosing an AK Treatment You'll Actually Finish

The trial that ranked them, and the arm that vanished

For decades, choosing between actinic keratosis treatments meant comparing trials that had never been run against each other. That changed in 2019, when the New England Journal of Medicine published a four-arm randomised trial from four Dutch hospitals. Patients with five or more lesions in one continuous area of the head were assigned to 5% fluorouracil cream, 5% imiquimod cream, methyl aminolevulinate photodynamic therapy, or 0.015% ingenol mebutate gel. The endpoint was demanding and sensible: at least a 75% reduction in lesion count, sustained to twelve months after treatment ended.

The cumulative probability of success was 74.7% for fluorouracil, 53.9% for imiquimod, 37.7% for MAL-PDT and 28.9% for ingenol mebutate. Fluorouracil beat all three, and not narrowly.

Then the story acquired a second act. Ingenol mebutate had been the convenience option — two or three days of application against fluorouracil's weeks, which is why it sold. In 2020 the European Medicines Agency's pharmacovigilance committee reviewed accumulating safety data and concluded the drug may increase the risk of skin malignancy; one comparison found 6.3% of ingenol-treated patients developing skin cancer in the treated area at three years against 2% treated with imiquimod. The marketing authorisation in the EU was withdrawn.

So the shortest, easiest, most convenient option turned out to be the least effective in a head-to-head trial and was subsequently pulled for possibly causing the disease it was prescribed to prevent. Hold that alongside every appealing new short-course option, including the ones in this article. It is not an argument that short courses are bad. It is an argument that convenience is a claim requiring the same evidentiary standard as efficacy, and that it is usually the last thing anyone checks. Our ↗actinic keratosis pathway tracks what is currently first-line.

Three mechanisms, three completely different bargains

These treatments are not variations on a theme. They kill abnormal cells by unrelated routes, and the route determines the experience.

Cryotherapy is destructive. Liquid nitrogen at −196 °C creates intracellular ice, which shears membranes, and the subsequent thaw damages the local microvasculature so the tissue dies from two directions. It is fast, needs no equipment beyond a canister, and is done in the room. Its defining limitation is that it treats what the clinician can see and chooses to freeze. It is a lesion treatment.

Fluorouracil is cytotoxic. It inhibits thymidylate synthase, starving cells of the thymidine they need to replicate DNA. Rapidly dividing dysplastic keratinocytes are hit hardest, which is why the cream selectively inflames abnormal skin and largely spares the normal skin beside it. That selectivity is the mechanism and the side effect: the angry map that appears on your face after a week is the drug finding lesions you could not see.

Imiquimod is immunological. It agonises toll-like receptor 7, prompting local production of interferon-alpha and other cytokines, and recruits the immune system to recognise and clear the abnormal cells. Nothing in the tube is directly cytotoxic — your own immune response does the work, which explains both its slower, more variable course and its capacity to make you feel systemically unwell.

The practical consequence is the field-versus-lesion distinction, and it is the most useful concept here. Chronically sun-damaged skin carries a field of genetically altered keratinocytes, most of which are not yet visible as lesions. Freezing the four you can see leaves the field intact. A cream applied across the whole area treats the field, which is why it hurts more, takes longer, and produces the more durable result.

Reading the numbers honestly

Every figure in this section comes with a condition attached, and the conditions matter more than the rankings.

The 74.7% for fluorouracil is a 12-month sustained 75%-reduction figure in patients with confluent facial and scalp disease — a hard endpoint in a demanding population. It is not "three quarters of people end up clear." It is "three quarters still had at least three quarters of their lesions gone a year later," which is genuinely impressive and not the same claim.

Cryotherapy's numbers are the most misunderstood, because they are less a property of the treatment than of the person holding the canister. A prospective study of 421 lesions in 90 patients found complete cure in 39% of lesions frozen for under five seconds, 69% for over five seconds, and 83% for over twenty. That is a treatment whose efficacy more than doubles based on how long the clinician holds the trigger — and holding it longer means more pain, more blistering, and a higher chance of a permanent hypopigmented dot. When cryotherapy underperforms, undertreatment is usually why.

Tirbanibulin, the newest topical, illustrates the trade with unusual clarity. In two phase 3 trials totalling 702 patients, five days of once-daily 1% ointment produced complete clearance at day 57 in 44% and 54% of patients against 5% and 13% for vehicle. Five days. But the same paper reports that at one year, an estimated 47% of those complete responders had recurrent lesions — and the authors explicitly note that trials comparing it with conventional treatments, with longer follow-up, are still needed. A five-day course is a real advance in tolerability. It is not yet demonstrated to be a five-day version of the four-week result.

And there is a benefit that sits outside all of these clearance tables. A randomised, double-blind, placebo-controlled trial published in JAMA Dermatology gave veterans a single course of 5% fluorouracil and followed them: in the first year, squamous cell carcinoma occurred in 1% of the fluorouracil group against 4% of controls — a 75% risk reduction. That is not lesion counting. That is a course of cream measurably reducing cancer incidence, and it is the strongest argument in this entire piece for tolerating the harder option.

What your face does, and why nobody warns you properly

The clinical literature reports "local skin reactions" on four-point scales. Here is the version people need before they start, because the gap between the two is where abandoned courses come from.

With fluorouracil, the first week is uneventful and mildly reassuring. Weeks two and three are not. Treated skin becomes red, then eroded, then crusted, and the pattern spreads well beyond what you thought was affected — a topographic map of damage you did not know you had. It stings, it weeps, it looks like a burn, and people will ask what happened to you. Peak reaction lands somewhere around weeks two to three, and resolution takes a further two to four weeks after stopping. Total: about six weeks from starting to looking normal.

Imiquimod is slower, patchier and less predictable, spread across weeks rather than concentrated. The distinguishing feature is systemic: because you are inducing an interferon response, flu-like symptoms — fatigue, aches, headache, sometimes fever — are a recognised effect and can arrive without much warning. That is a completely different planning problem from a red face, and our answer on ↗flu-like symptoms on imiquimod covers what is expected and what isn't. There are also situations where the choice is constrained rather than preferential — the answer on ↗imiquimod and breastfeeding is one.

Cryotherapy is minutes of real pain, then a blister, then a scab for one to two weeks per lesion. Localised, brief, and easy to conceal if there are only a few — which is precisely why it feels so much more manageable and why so many people end up having it repeatedly instead of treating the field once.

So the scheduling question is not squeamishness, it is planning. Six weeks with no wedding, no work photography, no holiday you have paid for. Winter is the obvious window: less UV on skin that is temporarily raw, fewer social obligations, and long sleeves and hats are unremarkable. Starting in May because that is when you noticed the lesions is how a course gets abandoned in week two.

Photograph it. The Photo Tracker is worth using here for two reasons — a same-light, same-angle series is the only reliable way to judge clearance across a treatment where the skin looks worse before it looks better, and seeing the reaction peak and then recede on a timeline is what stops people stopping. And take the ↗photographs before you start, because a baseline you didn't capture cannot be recovered.

Visits, cost and what your access actually allows

The logistics differ more than the efficacy does, and for many people they decide it.

Cryotherapy is a single appointment with immediate treatment, typically billed per lesion, and usually needs no follow-up beyond a check. If your access to dermatology is one appointment after a long wait, this is what will happen at that appointment, and it is a reasonable thing to happen.

Topical field treatment inverts that. One consultation, one prescription, and then several weeks of unsupervised self-administration at home. Drug cost is modest, particularly for generic fluorouracil. What it demands instead is the ability to keep applying something that is visibly making your skin worse without anyone checking on you — which is exactly why a planned review call at the two-week mark, when the reaction peaks, changes completion rates. Ask for one.

Photodynamic therapy is the most clinic-intensive: application, an incubation period, then illumination, with the light exposure itself frequently described as burning or stinging. It clears in days rather than weeks and concentrates all the unpleasantness into a single supervised session, which suits some people very well and is unavailable or expensive in many places.

Before that appointment, being specific pays. The Derm Visit Prep tool assembles the things that change the recommendation: how many lesions and where, what has been treated before and how it went, whether anything is thickened, tender or growing, and — the input people never volunteer — what the next two months of your life look like. The Skin Cancer Risk tool is worth running alongside it, because cumulative risk is what determines whether this is a one-off tidy-up or the beginning of ongoing surveillance.

Combining, sequencing, and what to actually do

These are not mutually exclusive, and the sensible plan usually uses more than one.

The common and rational sequence is destructive treatment for the outliers plus field treatment for everything else: freeze the thick, hyperkeratotic lesions that a cream will struggle to penetrate, then run a topical course across the region to address the field they emerged from. Doing the field treatment first also has diagnostic value — anything that refuses to clear after a proper course has declared itself as needing biopsy rather than another cream. A lesion that survives adequate treatment is information, not a reason to repeat the treatment.

The supportive care is dull and makes a real difference to completion. Wash with something that does nothing: a plain, non-foaming cleanser such as Vanicream Gentle Facial Cleanser, with no acids, no scrubs and no actives of any kind for the duration. Once erosion and crusting begin, a bland occlusive like CeraVe Healing Ointment applied between doses reduces the discomfort meaningfully without blunting the treatment. Nothing else goes on your face until it has healed.

And then the part that determines whether you are back in three years. Field cancerisation is cumulative UV damage, and treatment resets the visible lesions without resetting the exposure. Daily high-protection sunscreen on the scalp, ears, nose and lip margin — the sites that get treated and retreated — is the only intervention that changes the trajectory rather than the current state. A robust mineral formula such as Blue Lizard Sport Mineral Sunscreen SPF 50+ suits skin that is still recovering, and something like La Roche-Posay Anthelios Melt-In Milk Sunscreen SPF 100 covers scalp and body for people who are outdoors for a living. A hat outperforms both.

The recommendation, stated as plainly as the evidence allows: if you have a field of damage rather than three discrete spots, and you can find six weeks in your calendar where looking alarming is survivable, fluorouracil is the option with the best twelve-month data and the only one with randomised evidence of reducing subsequent squamous cell carcinoma. If those six weeks genuinely do not exist — and for plenty of people they don't — then imiquimod's slower, more diffuse course or a five-day tirbanibulin course is a legitimate second choice, provided you understand you are trading durability for tolerability and should expect to be back. What is not a plan is repeat cryotherapy of whatever is visible, every eighteen months, forever, on a field nobody has treated.