The Gerson Protocol: Juices

Potassium, ASS1, and the fragment of a protocol worth testing

Three Glasses, Not Thirteen

Many people who try the Gerson protocol only adopt the juicing, and even then, almost no one manages the prescribed dose. That makes the real-world exposure a different intervention from the one that has been argued about for seventy years—and the only version with a mechanism behind it. Just not the mechanism Gerson gave.

A previous post argued that Max Gerson’s dietary reasoning fails at a single point: the sodium–potassium gradient he tried to refill is not a reservoir but a pump, and nothing on a plate reaches it. That argument stands. It is also incomplete, because it answers a question almost nobody is actually asking.

Hardly anyone does the entire protocol. What people do is drink vegetable juice, and far less of it than the regimen specifies. That is a different exposure with a different risk profile, and it deserves to be evaluated on its own rather than inherited from a regimen many of its users have already discarded.

Potassium depletion suppresses ASS1 transcription and cuts urea synthesis capacity by a third, with an eightfold rise in ammonia—all of it reversed by repletion. That makes a potassium-rich diet a plausible lever on nitrogen handling in a depleted patient. It is not Gerson’s mechanism, it is not a supply effect, and it has never been tested in a human being.

Part I / The fragment that survives

What People Actually Do

“Gerson therapy” has no fixed exposure. The published regimen runs to thirteen juices a day, potassium salt solution, thyroid extract, pancreatin, Lugol’s solution, up to five coffee enemas and—until 1989—injected raw calf liver. The version that circulates is the first item on that list, usually alone, usually at a fraction of the dose, and alongside ordinary food rather than instead of it.

This is normally treated as a nuisance for the analyst: no fixed exposure means nothing to test. It is better read as information. The components people keep are the ones that are cheap, tolerable and compatible with eating normally; the components people drop are the ones carrying every documented death. The surviving fragment is not a watered-down protocol. It is a different intervention that happens to share a name.

And it is the only part of the protocol with a defensible mechanism behind it — which is worth establishing carefully, because the mechanism is not the one on the label.

Part II / The mechanism he gave, and the one he missed

Right Organ, Wrong Compartment

Gerson’s claim was that diseased cells lose potassium and take up sodium, and that supplying potassium would restore the intracellular balance. The first half is a real observation, confirmed decades later by ²³Na-MRI. The second half cannot work: intracellular potassium is set by the Na⁺/K⁺-ATPase running continuously against the gradient, and plasma potassium is held in a narrow range by the kidney whatever arrives at the mouth.

But there is a second route from dietary potassium to the disease, and Gerson never mentioned it, because the enzyme at the centre of it was not characterised in his lifetime. It runs through the liver, and it operates on gene expression rather than on ion concentration.

Part III / Where the brake sits

A Third of the Cycle’s Capacity

Mikkelsen and colleagues put rats on a potassium-free diet for thirteen days, reaching a plasma potassium of 1.9 mmol/L, and measured the capacity for urea-nitrogen synthesis directly in vivo.^[1] Capacity fell by 34%. Plasma ammonia rose roughly eightfold, from 28 to 235 µmol/L. Potassium repletion reversed all of it.

The comparator matters: pair-fed controls ate the same amount, so this is the effect of potassium itself rather than of eating less. ASS1 was the only urea cycle enzyme that moved against that comparator.

Two details make this more interesting than a general sick-animal effect.

The first is specificity. Against pair-fed controls—animals eating the same restricted amount—only ASS1 changed, falling 33%. OTC and arginase did not move. CPS1 and ASL dropped against free-fed controls but not significantly against pair-fed ones, which is what you would expect from reduced intake rather than from potassium. The lesion is narrow, and it sits at a single step.

The second is the tier it occupies. In Where the Cycle Broke I set out four levels at which a urea cycle can be suppressed—transcription, protein activity, allosteric regulation, substrate supply—and argued that the level determines what, if anything, can be done about it. This is the top rung: a transcriptional brake. It is also the rung with the best prognosis, because transcription is reversible and the repletion arm demonstrates the reversal.

A suppressed cycle with an identified, correctable cause is a rare thing. Most of the mechanisms that shut down urea cycle enzymes in cancer—promoter methylation, HIF-driven transcriptional programmes, outright deletion—are not addressable from the outside. A potassium deficit is.

Part IV / A deficit, not a supply

Who This Could Possibly Help

Here the earlier argument has to be applied symmetrically, because it cuts in both directions. A quantity held at a set point by active machinery cannot be pushed above that point by intake—which is why Gerson’s version fails. But the same quantity can fall below it, and when it does the consequences are real and repletion is effective. Correcting a deficit and raising a normal value are different operations, and only one of them is available.

So the question is not whether potassium is good for you. It is who is actually depleted.

Depletion is common in cancer

Hypokalemia affects around 12% of cancer patients overall and 43–64% in acute leukemia, driven by renal potassium wasting, treatment-induced diarrhoea and tubular injury.^[2]

Magnesium is usually the engine

Hypomagnesemia reaches 90% with cisplatin.^[2] Low intracellular magnesium releases its block on ROMK channels, so the distal nephron leaks potassium continuously — which is why such hypokalemia is refractory to potassium alone.^[3]

Baseline intake is already short

US adults average 2,937 mg (men) and 2,324 mg (women) against Adequate Intakes of 3,400 and 2,600 mg; roughly three-quarters fall below.^[4]

That third card is the one that makes the modest version interesting. A cup of carrot juice carries about 689 mg of potassium, tomato juice 527 mg, orange juice 496 mg.^[5] Two or three glasses a day closes a typical shortfall. Thirteen glasses delivers somewhere near 9,000 mg, which is not a larger dose of the same thing — it is a different problem, particularly in anyone whose renal potassium handling is impaired.

Component	The published protocol	What many people do	What the evidence actually addresses
Juice volume	13 glasses daily, roughly 9,000 mg K⁺	1–3 glasses, roughly 700–2,000 mg	Closing a deficit. There is no evidence that exceeding adequate intake adds anything
Protein	Restricted; animal protein excluded initially	Normal diet alongside the juice	1.2–1.5 g/kg/day is current guidance in cirrhosis; restriction was abandoned[6]
Fibre	Removed by juicing and by milling the soup	Removed by juicing	Fermentable fibre acidifies the colon and traps NH₄⁺ — the lactulose mechanism[7]
Coffee enemas	Up to five daily	Usually omitted	Cause the potassium loss the juice is meant to correct; the reported deaths were electrolyte derangement at autopsy

Read down the last column and the protocol argues against itself. The component most often dropped is the one that creates the deficit the retained component would repair.

Part V / The fraction he removed

What the Protocol Throws Away

Two of the mechanisms that would help most are damaged by the preparation rather than by the ingredient. In one case the damage is deliberate, and it is applied across the whole regimen rather than being an artefact of juicing.

Fibre. An established intervention for hyperammonemia is lactulose, which works by acidifying the colon so that ammonia is protonated to ammonium and cannot cross the mucosa. Fermentable fibre does the same thing less powerfully, and adds faecal bulk that carries nitrogen out of the body. Juice contains almost none of it.

The obvious objection is that the protocol is not only juice. It also specifies Hippocrates soup — celeriac, parsley root, leeks, onions, potatoes, tomatoes and garlic, unpeeled, simmered for hours and eaten at both lunch and dinner.^[9] On the face of it, that is a substantial daily dose of vegetable fibre sitting right there in the regimen.

Except that it is not eaten that way. The instruction is to pass the soup through a food mill before serving; in the stock-based version the simmered vegetables are strained out and discarded outright. A food mill is not a fine sieve and some soft pulp passes, so the soup retains more than a juice does — but the coarse insoluble fraction is removed by design, at the one point in the protocol where it would otherwise have arrived intact.

And the partition that straining produces runs exactly the wrong way. Potassium is water-soluble and leaches into the cooking liquid, so straining keeps it — this is why renal patients are told to boil vegetables and discard the water. Fibre is insoluble and stays in the residue, so straining removes it. The preparation retains the component that cannot work, because plasma potassium is defended by the kidney, and discards the component that could, because colonic acidification is defended by nothing.

Gerson’s reason for removing the residue was that a burdened system should not have to digest. It is the same instinct as the juice and the same instinct as the enemas: reduce what the body is asked to process. Applied to fibre, it strips out the fraction with the strongest claim on the mechanism his diet actually needed.

Protein. Skeletal muscle is the principal extrahepatic ammonia sink, clearing it through glutamine synthetase, and it is what holds the line when hepatic capacity falls. A low-protein diet shrinks that buffer. Protein restriction in hepatic encephalopathy was abandoned for exactly this reason: the sarcopenia costs more than the nitrogen load saves.^[6] The full Gerson diet is protein-inadequate by construction, so taken as prescribed it works against the mechanism that gives it its only plausible rationale.

The protocol is the worst available version of its own idea. At thirteen glasses with protein restricted and the colon flushed, every supporting mechanism is cancelled and the hazards are maximal. At two or three glasses alongside a normal diet, the mechanism is intact and there is no meaningful hazard. The dose–response here runs backwards from the way it is usually argued.

Part VI / The complication

The Same Gene, Two Tissues, Opposite Wishes

There is a difficulty that has to be stated, because it is the kind that gets left out of enthusiastic accounts.

ASS1 is not only a hepatic urea cycle enzyme. It is one of the better-studied genes in tumour metabolism, and it is silenced in a wide range of cancers—which is the basis of arginine-deprivation therapy. Tumours that cannot make their own arginine die when it is withdrawn, and pegylated arginine deiminase exploits exactly that. Cisplatin suppresses tumour ASS1 through a DEC1–HIF-1α–MYC network, and that suppression is what creates synthetic lethality to arginine starvation.^[8]

So the gene we would want up in hepatocytes for nitrogen clearance is the gene we may want down in tumour cells for therapeutic vulnerability. Whether a systemic signal such as potassium status acts on both compartments in the same direction is unknown, and it is not a question that can be waved away. Anyone on arginine-deprivation therapy is a specific reason for caution rather than a general one.

Part VII / The disguise again

The Liver’s Potassium Never Changed

The detail that should unsettle anyone who wants to monitor this is buried in the tissue measurements. Kidney potassium fell 17%. Muscle potassium fell 27%, and muscle sodium rose—incidentally reproducing, in a potassium-deficient animal, the tissue signature Gerson described in a diseased one. Liver potassium did not change at all, and hepatic Na⁺/K⁺-ATPase protein was unchanged with it.

The liver defended its own potassium successfully and lost a third of its urea cycle capacity anyway. The suppression is not a local substrate effect; it is a signal arriving from a systemic state. Which means the organ doing the failing shows no abnormality in the quantity that caused it.

This is the same structure as Ammonia in Disguise: the number you would think to measure sits in its normal range while the process it governs is substantially impaired. A normal hepatic potassium—or for that matter a plasma potassium at the low end of reference—does not rule this out.

Part VIII / The missing experiment

What Would Settle It

Everything above is one rodent study, a set of epidemiological frequencies and a mechanism. The authors of that study closed by calling for clinical work. Five years on, no human trial has measured urea synthesis capacity before and after potassium repletion, in liver disease or in cancer.

The experiment is not difficult and does not require a new agent. Take patients with documented hypokalemia, correct it properly—meaning magnesium first, since otherwise the kidney keeps leaking—and measure the capacity for urea-nitrogen synthesis, plasma ammonia and glutamine before and after. The question is whether the rat finding transfers, and the method for answering it has existed for decades.

The mechanismEstablished in rodents

Potassium deficiency suppresses ASS1 and cuts urea synthesis capacity by a third, reversibly

The populationIdentifiable

Depleted patients, not everyone — and magnesium has to be corrected alongside

The human evidenceAbsent

No trial has measured urea synthesis capacity before and after repletion in any patient group

A mechanism without a clinical test is a hypothesis, not a treatment. The reason to state it precisely is so that it can be checked rather than believed.

Part IX / The transferable lesson

Dose Is Part of the Claim

The habit this piece is arguing against is one both camps share: treating a protocol as a single object that is either endorsed or dismissed. Gerson’s defenders inherit the mechanism along with the juice. His critics inherit the enema deaths along with the vegetables. Neither group is evaluating the thing that many people are actually doing instead.

A component has a dose, a format, a mechanism and a population. Change any one of them and it is a different claim requiring different evidence. Thirteen glasses of juice with protein restricted is not a strong version of two glasses with dinner; on the available mechanism it is a worse one. That is not a defence of Gerson’s theory, which remains wrong in the way the earlier piece set out. It is an argument that the question worth asking was never the one being fought over.

He prescribed the right food for the wrong organ, at a dose that undid the only benefit it could have had, and milled out the part that might have helped most. The interesting part is that a correctable version of his idea has been sitting in the literature since 2021, and nobody has taken it to a patient.

References

Sources are listed in order of first citation. The notes identify the kind of evidence each contributes.

1. Mikkelsen ACD, Thomsen KL, Vilstrup H, et al. (2021). Potassium deficiency decreases the capacity for urea synthesis and markedly increases ammonia in rats. American Journal of Physiology: Gastrointestinal and Liver Physiology 320(4):G474–G483.The core finding; rat model with pair-fed controls and a repletion arm ↩
2. Electrolyte disorders induced by antineoplastic drugs (2020). Frontiers in Oncology.Prevalence figures for hypokalemia and hypomagnesemia in cancer populations ↩
3. Huang CL, Kuo E (2007). Mechanism of hypokalemia in magnesium deficiency. Journal of the American Society of Nephrology 18(10):2649–2652.Why magnesium must be corrected first: loss of intracellular Mg²⁺ block on ROMK ↩
4. Potassium intakes of the US population, NHANES 2017–2018. USDA Agricultural Research Service, Dietary Data Brief 47.Mean intakes against the NASEM Adequate Intake values ↩
5. Food sources of potassium, standard portions. Dietary Guidelines for Americans.Potassium per cup for carrot, tomato, vegetable and orange juice ↩
6. EASL Clinical Practice Guidelines on nutrition in chronic liver disease. Journal of Hepatology.Protein targets of 1.2–1.5 g/kg/day and the case against restriction ↩
7. A comprehensive review evaluating the impact of protein source (vegetarian vs. meat based) in hepatic encephalopathy (2021). Nutrients 13(2):370.Fibre, colonic acidification and nitrogen excretion; notes the clinical evidence is weak ↩
8. Cisplatin-induced synthetic lethality to arginine-starvation therapy by transcriptional suppression of ASS1 is regulated by DEC1, HIF-1α, and c-Myc transcription network. Oncotarget.The opposing interest: tumour ASS1 suppression as a therapeutic target ↩
9. Hippocrates’ soup. Gerson protocol recipe documentation.Cited as a primary source for what the regimen specifies — ingredients, the food mill, and discarding the strained vegetables — not for any claim made about it ↩

Evidence note. Targeted primary-literature search through 20 October 2026. This article is mechanism analysis, not treatment guidance. The central finding is from a rodent study and has not been tested in humans; nothing here establishes that drinking vegetable juice treats cancer, and it does not. Potassium supplementation is hazardous in renal impairment, in heart failure, and alongside potassium-sparing diuretics or ACE inhibitors, and hypokalemia driven by magnesium loss will not correct without magnesium. Anyone considering a change in potassium intake during cancer treatment should raise it with their oncologist first, and anyone receiving arginine-deprivation therapy has a specific reason to do so.