Gary Curhan and Eric Taylor have given us many insights into how diet might influence kidney stone risk. I think this new article by them and their colleagues a great opportunity for close reading and practical use of a high quality research paper remarkably germane to the practice of kidney stone prevention.
What They Want
We all know that science is about discovery that enables us to do something, or know how nature does something. As a consequence, most of us ask what kind of science a paper is about – doing or knowing. But we also all know that this dichotomy is false because almost every scientist is after at least some of both. Likewise, that a paper lies between the two poles of doing and knowing – that, too is false. Far better to say any fine research has two separate gauges of new discovery: one about doing and one about knowing.
What about this one?
Do Something Research
Certainly diet is a massive factor in stone formation and prevention. Protein is a main player because it raises urine calcium excretion, some say because cystine and methionine are metabolized to acid. Likewise for potassium in foods. Whether in plants or animals, potassium’s counter ion is not chloride, but rather anions that we metabolize to bicarbonate. So foods rich in potassium can act like potassium citrate pills, to put things in a crude if direct way.
Alkali loads lower urine calcium and raise urine citrate – good for preventing calcium stones. Acid loads the exact opposite.
Given these antinomies, one wants to know how food protein and potassium affect stone formation, so we can better use diet in stone prevention. Plants usually have an overabundance of potassium alkali over acid producing amino acids. Food protein of animal origin also has potassium – cells store it, but acid production from amino acids usually overbalances alkali production from potassium anions. So this class of food provides not an alkali but an acid load – the opposite of potassium citrate pills.
That is not all. Dairy foods come from animals but unlike meat have large amounts of calcium that can lower urine oxalate and reduce stone risk.
Since we can do something about diet, this certainly is ‘do something’ research. Given their differing proportions of acid and alkali generation, these three food sources may indeed affect stones and offer a chance to reduce stone risk by altering their relative proportions in the diet.
Know Something Research
The investigators also want to know something. All agree that protein metabolism gives rise to an acid load, or seems to, and likewise agree that acid loads increase urine calcium, lower urine citrate, and therefore raise kidney stone risk that varies with both. But not everyone agrees about protein and urine calcium. Some find that protein feeding raises urine calcium even when it imposes no acid load at all.
If protein raises urine calcium independent of acid load, potassium alkali from plant sources would not directly null its effects on stone risk, but only do so indirectly by lowering urine calcium and raising urine citrate from the action of alkali.
Perhaps, they reason, the balance between diet potassium – an alkali source, and diet protein – an acid source, might itself influence stone formation. If so, that would bolster the ‘protein as acid’ school of thought concerning stone risk. Although as we state the proposition one can see some difficulties with this test of the acid theory.
What They Already Know
Prudent scientists make sure they know what has already been discovered before setting out to discover something new. The authors are mandarin scholars.
The Main Facts
Curhan and Taylor – the architects of this work – do skilled epidemiology using three well established cohorts of women and men followed, by now, over decades. From their own studies they know that protein intake per se, not separated as to source, has a weak and variable association with becoming a stone former. Likewise, they know that increasing food potassium intake associated with progressive fall in risk of stones, but only in two of their three cohorts – so it has seemed not as vigorous an association as some others – urine calcium, oxalate, and citrate excretions, as examples, and urine volume itself.
They imagine that diet protein raises stone risk by imposing an acid load that can raise urine calcium, lower urine citrate and perhaps cause other effects as yet to be defined. When evaluated by source, animal protein will most strongly associate with risk of stones because it imposes the largest acid load. Vegetable proteins offer very much less acid load, and dairy proteins somewhere in between.
They also imagine that with more observations, such as they now have, the potassium effect will be significant in all three cohorts because potassium alkali must raise urine citrate, a powerful defense against stone formation.
Because acid load is envisioned as stone producing and alkali load the opposite, they envision that the play between intakes of animal protein and potassium should strongly influence stone risk via modulation of net acid load.
From this they must have deduced three necessary outcomes were one able to somehow assess effects of proteins and potassium on risk of kidney stones:
- Animal protein will associate with increased stone risk, vegetable protein will not, and dairy protein perhaps somewhere in between.
- With more observations, diet potassium will strongly associate with reduced stone risk
- The relationship between animal protein and potassium intake – the ratio of the two as they propose to test the idea – will strongly associate with stone risk.
What They Did
They did what they always do: Highly sophisticated epidemiology.
They have information from three cohorts, two of women (NHS I and II, nurses) and one of men (HPFS, physicians) followed for many years. Among them, some became kidney stone formers, most did not. All had filled out elaborate and well calibrated food diaries at least once before forming any stones. As well, the scientists collected 24 hour samples from the stone formers and a properly selected parallel set of well matched people who never formed a stone. From comparisons of diet intake and urine values between the two groups, and using a highly elaborated mathematical technique they extract the relative risk of becoming a stone former from various intake levels of the three protein sources.
Elsewhere on this site I have presented their work showing the effects of urine calcium, oxalate, volume, and citrate on this relative risk.
What They Found
Like most professional scientists, this group presents their answers in dense tables from which I have extracted and graphed what seem to be the key results related to their deduced outcomes.
Each bar shows the the mean relative risk at its top, and at its bottom the lower 95th percentile for that risk. For clarity, values above 1 (the dashed horizontal line) are lighter, those below deep red.
Contrary to prediction, risk does not vary significantly with animal protein intake. I show here the pooled values for all three cohorts, but results are essentially the same for each one taken separately. At the highest quartile of animal protein intake the bottom 95th percentile does not go below 1, so the relative risk is significantly increased. But increasing amounts of animal protein do not create increasing risk of stones (p=0.2).
I plot their multivariate corrected pooled values adjusted for BMI, presence of diabetes or hypertension, use of thiazide diuretics or supplemental calcium, and for intakes of fluid, sodium, potassium, fructose, oxalate, phytate, alcohol and all other types of protein.
As they do not give intake quartiles for pooled values, but values in the three cohorts differed only by small amounts, I used those for the NHS II cohort because they lay closest to the bottom of the table above the pooled values.
Vegetable protein had no effects, so I omit graphing them from the supplemental tables.
Because in general this form of protein lowered risk slightly, I plot the mean risk ratio at the bottom of each lighter bar and the upper 95th percentile at the top.
The highest quartile of intake has an upper 95th percentile below 1 and therefore is significantly protective, but overall as a trend, rising amounts of dairy protein did not confer progressive amounts of protection as is obvious from the graph (p=0.06 for the trend).
As in the prior graph I have plotted the fully adjusted values from the pool of the three cohorts. This time I used diet intake quartiles from the male HPFS cohort – for variety. Note how much smaller are intakes of dairy than animal protein – about 4 fold.
Their Protein Hypothesis
Falsified by Their Data
Their experiment falsifies a necessary prediction of their hypothesis that protein intake – especially animal protein intake – raises kidney stone risk. None of the three protein sources had a robust effect on stone risk. Within the actual ranges that dominate the populations they studied that effect cannot be demonstrated.
When Is No Really No?
Karl Popper founded the philosophical groundwork for ‘no’ as the only sure guide to a form of reliable truth. Put simply, and poorly, a hypothetical idea cannot be universally true if even one of its necessary predictions proves false. But this leaves open the question of domain – can an idea be useful within one domain yet fail elsewhere?
Great Protein Excess May Pose Risk
The very highest animal protein load did seem to have some baleful effect. So massive protein loading from concentrates, or even from very large meat intakes, could pose risk. They did not specifically set out to test that idea, and it may prove false if tested with more vigor.
The Common Intakes Create No Risk Gradient
But over the ambient protein intake ranges of people in numbers and over time they detected no effect of protein on stone risk. So their ‘no’ pertains to a specific but large and practical domain, and also to the techniques and methods they used to measure effects.
Falsification Has Practical Value
Because among people in numbers and over time protein intake scarcely alters stone risk, in the commonplace business of stone prevention protein intake need no longer attract our interest. This is a powerful statement for physicians to make to their patients. It is one burden to lift from them, one complexity they can ignore, for the most part, in perfect safety. We need only caution against excesses, and even there we have no proof as yet.
Does Not Test the Acid Load Hypothesis
The negative result does not invalidate the idea that protein can promote stones via acid load. But it does demonstrate that the acid loading from actual ranges of protein intake are not sufficient to vary kidney stone risk. In other words, they have tested the relevance of protein under common conditions in relation to stone risk, but not the underlying mechanism of acid load promoting stones as a basic physiology.
Because women and men differ in size, I cannot plot them over one set of intakes, so the two female cohorts (red) are on the left, the male cohort (blue) on the right of this two plot. For female quartiles I chose those of NHS II.
I converted all intakes into mEq/d because this site has preferred that unit. One multiplies by 39 (the atomic weight of potassium) to get household units of mg/d.
As potassium intake rises, relative risk of becoming a stone former falls markedly and progressively in both sexes, and the upper 95th percentile of the risk is far below 1 (upper ends of the lighter bars). Compared to the reference ranges, below 63 and below 72 mEq/d, women and men, respectively, the protective effects are evident and significant in the first quartiles, with the exception of one of the female groups (red bar extends above 1).
Their Potassium Hypothesis
Quite opposite to protein, this food component has a powerful effect, and therefore physicians can have confidence in recommending high potassium foods to their stone patients. Because some have significant oxalate contents, we need to choose with some care and use diet calcium as an offset. Several articles have detailed this approach. More generally, the kidney stone diet, like the US recommended diet, is a whole food plan that includes high diet calcium, reduced sodium, and ample servings of fruits and vegetables. The latter provide much of the potassium load.
When is Yes Really Yes?
In a way, forever. These data will hold, no doubt indefinitely, as will the negative data concerning protein. Others may find differently among different people, in different places, with different techniques. But those who repeat what Curhan and Taylor did will almost certainly find what they found. If not, will ensue the scientific homologue of disagreement as to simple facts, thus calling for more repeated work that inevitably discloses some final consensus.
But what will inevitably fall is the idea underlying this study, that alkali protect against stones, acids promote stones, and the one can offset the other – in other words, the mechanistic vision this work arises from. Given the history of science, these mechanistic visions of how nature does things rarely last but are – as it were – overlain by revelation presently unthought of.
Even so, while we wait for their inevitable demise, these two formulations have met various tests and remain viable explanations for what we can observe. This makes them useful, still, as generators of new research.
Correlates of Diet Potassium
As a kind of gift, the data of excellent science often enough give us useful or at least eye opening glimpses of how things work. Here are four urine measurements central in kidney stone diagnostic evaluation plotted as a function of diet potassium intake – that means veggies and fruits for the most part. The authors tell us that stone formers and controls scarcely differed, so they offer the pooled values.
In all four plots, the big blue dots are medians, the lines the 95% confidence limits. All four trends were significant at p<0.001.
Indeed, as diet estimates show higher values, actual urine potassium rises in excellent accord. The excellent correlation makes clear how well the diet estimates gauge potassium intake. The red line of identity lies far above, as only a fraction of ingested potassium is absorbed and lost in the urine.
Urine Citrate and pH
Food potassium is mainly as the salts of anions that can be metabolized to bicarbonate – put crudely. This should raise urine citrate and pH.
Urine citrate, rises in a haphazard and rather sluggish way, from about 675 to 710 mg/d over the wide range of potassium intakes. This is true even if we take the lower 95% boundaries of the confidence intervals. The three cohorts had high average citrate excretions, as it were, and the effect of potassium intake is small. Therefore the expected protective effects of citrate from diet potassium would seem likewise small.
The latter rises in fine progression with potassium intake, and over the range in which calcium oxalate stones will predominate. Even the outer boundary of pH for the highest pentile of diet potassium is below the region where calcium phosphate will predominate in stones. I view this as important. We can tell patients to eat their fruits and veggies without fear of provoking calcium phosphate stones as a general rule. Of course people do vary, and prudent physicians measure to be sure.
This rose with more vigor – to my eye – than did urine citrate, presumably because fruits and veggies contain a lot of water. The CaOx supersaturation fell (5.8 to 5.1 over the whole range of intakes) with rising potassium intake, perhaps because of this volume increase, and that may well explain at least part of the marked stone risk reduction from increasing potassium intake. Although the change in supersaturation does not seem so marked, we have as yet no data from this group about its effects on stone risk, which may be very powerful indeed.
Animal Protein to Potassium Ratio
Even though animal protein had no effect on stone risk, but potassium intake had a powerful effect, one might imagine that when the latter was low and the former high one might see an effect of the ratio even after adjusting for the total intakes of both. Put another way, the combination of high animal protein and low potassium intakes might foster stones even when total intake of one (animal protein) was itself not a driver of stone risk because the other (potassium intake) was.
The reason for suggesting this rather counter intuitive idea is not the obvious and rather debatable one that given no effect of the numerator (animal protein intake) and a powerful effect of the denominator (potassium intake) the ratio must show a strong relationship to stone risk. It is, I suspect, that individuals will vary in this ratio so that those with high ratios will become stone formers even if one corrects for total intakes of both animal protein and potassium – as they did.
Indeed, this prediction proved true for the males and for one of the two female cohorts. The pooled results for all three cohorts likewise showed a significant trend of increasing risk with rising ratios. For those with sufficient technical insight I recommend consulting the table of data in the original paper, and giving the matter some thought.
Clinically, I cannot see how this will help patients overly, as we want high diet potassium intake and that will, when successful, place most of the ratio effects on just animal protein intake, which latter has little effect by itself. To ask patients both a reduction of animal protein and increase of diet potassium seems excessive once we know the former has no independent effect. Every thing we ask of patients by way of change takes energy from them, and we need to focus on where risk most varies – in this case diet potassium in a spectacular manner.
Obviously this ratio derives it intellectual force from the food acid – alkali antinomy. As plant alkali – potassium – neutralizes animal acid, meaning as the ratio falls, the theory predicts a fall in stone risk, which was found. But, although this is a necessary prediction of the acid base theory, a positive result has enough alternative explanations to make it a weak confirmation.
If animal protein acid were, in fact, irrelevant, as some have said, and if potassium alkali were in fact a powerful protection via increase of urine citrate and reduction of urine calcium, those who chose a low potassium intake would have increased stone risk whether or not animal protein acid had any independent effect.
So the acid – alkali formulation has passed a necessary test and was not falsified, but the test lies far shy of support for the animal acid portion. This is especially true because protein whose entire animal acid load has been neutralized with alkali still raises urine calcium.
Thus does debate over theory inevitably roil one more than disagreement over fact. This group always has its facts right, and no group ever gets its theories accepted but by mighty effort.
What Do We Have For Our Patients?
Eat Your Fruits and Veggies
As usual with results from the Curhan/Taylor group we have something very excellent. Diet potassium increase seems an unmitigated benefit and of great value. The upper end of the range, about 100 mEq/d, or about 4,000 mg/d, is at the US recommended ideal. So this research about stone disease more or less coincides with the massive research upon which a generation of scientists have based diet recommendations for the whole of the American people.
Apart from massive support for what everyone’s mothers have long urged, we have a crude yet tolerable estimator of diet intake from urine potassium. More or less, an ideal intake should give us somewhat over 70 mEq/d of urine potassium, and from their data as I have graphed it, one can estimate – albeit most crudely – about where a given patient might lie in terms of eating.
Past Fears of Animal Protein Seem Bogus
No one questions that massive protein excess can raise urine calcium. But either way, from acid or not, within the range of large numbers of people over long time periods these investigators find no support for animal protein intake per se as a cause of stones. That dairy proteins may have minimal protective effects could be related to the calcium that accompanies it, or perhaps other factors in milk products; but the effects are small. So, enough protein – 0.8 gm/kg/d and not too much – no more than 1 gm/kg/d seems as prudent for stone formers as for the rest of the US, which matches the general US recommendation for us all. As for the protein in milk products, it is not any risk at all.
Dr Eric Taylor kindly read this article, at my request, and accepts my account as accurate. I thank him personally for going to this effort. Since his reading, I have slightly modified the logic of the first several sections with regards to their testing of the acid theory of protein stone genesis, but the main lines of this review remain unaltered.