Unlike Zeus, or Athene, Janus did not come to Rome from Greece but from myths about a person living early in Roman history and later deified. Janus – deity – presides over beginnings and endings, gateways and doors, invariably dual in nature.
What is dual here?
Calcium stone formers are dual. A minority arise from systemic diseases we must screen for. Each systemic disease has its own universe of causes and treatment decisions. A majority are “idiopathic”, systemic causes have been excluded.
Idiopathic calcium phosphate stone formers are dual. Most have hydroxyapatite (HA, like bone mineral) as their stone calcium phosphate. Some have brushite (Br, calcium monohydrogen phosphate) in their stones. These latter have more kidney damage than HA CaP stone formers, and are a special high risk group of patients.
Both kinds of CaP stone formers need special attention. That is why I have written this article for them.
Basic Facts about Phosphate Stone Formers
Phosphate Stones Damage Kidneys
Phosphate stones, HA or Br, can grow faster and larger than calcium oxalate ones. Calcium phosphate crystals invade kidney tissue – so called tubule plugs. Tissue damage is common, as is Nephrocalcinosis from plugging – often misdiagnosed as medullary sponge kidney. Kidney tissue damage is worse with Br than HA stones. Potassium citrate, a common stone prevention, may not be appropriate as a treatment because it raises urine pH.
Alkaline Urine Causes Phosphate Stones
Stone phosphate replaces oxalate when urine is too alkaline. Kidney and GI tract physiology raise urine pH, especially in women. Diet is not the cause of higher urine pH. Diet will not reliably lower the pH, and we have no specific drugs to do it, either. So although treatment uses the same tactics as for the more common calcium oxalate patient, it must follow a different strategy.
How Stone Analysis Distinguishes CaP from CaOx Stone Formers
Only CaOx and HA Present
If the average stone mineral composition of all available stones for a given patient is above 50% calcium oxalate, the patient is considered a calcium oxalate (CaOx) stone former. If the average calcium phosphate content is above 50% the patient is considered a calcium phosphate stone former.
The average must be computed using 0 – for example, given CaOx/CaP percentages of 100/0, 0/100, 40/60, the correct classification is 140/3 vs 160/3 or 46% CaOx vs 53%, and so a CaP stone former.
Uric Acid, Struvite, Cystine Also Present
If uric acid, struvite, or cystine are present we name the patient for that constituent. A patient who forms mixed stones – for example, 60% calcium phosphate/20% struvite 20% CaOx is called a struvite stone former. The reason is that these stone types have special causes and treatments.
Any Brushite Present
Brushite is very uncommon in human kidney stones, and associates with large tubule plugs and more severe tissue damage. So when any stone contains brushite we classify the patient as a brushite stone former even though brushite is a minority of stone mineral.
Sex and Age
Single Clinic Experience
Percentages of Cases
The table shows ‘CaP’ as cases where HA or brushite was the stone phosphate crystal (in early years we did not distinguish). CaP(b) are CaP stone formers with only HA, no brushite in any stone.
CaP predominate among females (a and b superscripts denote outsize high frequencies). Brushite does not show this difference a statistical level of significance. CaOx stone formers predominate among the total of all cases. Brushite patients are least common.
Sex vs. Percent CaP in Stones
The same study furnished this nice graph showing the sexes as the percent of stone CaP increases. The bulk of patients have very little CaP in stones (tall bars at the left of the graph). These are the common CaOx stone formers, mainly men (% female, dots, right axis about 25%). But when CaP percent is 20 – 50% in stones, women and men are nearly equal.
This graph blurs the sex distinction because we used stone CaP% from both brushite and hydroxyapatite. Today, I would have left the brushite to one side, which would have made the female preponderance among those with high stone CaP% more marked – because the sex ratio for brushite stone formers is closer to 1.
National Laboratory Findings
The Mayo Clinic kidney stone analysis laboratory analyzed 48,446 stones in 2010, and of these 43,545 were the first submitted to the lab for that person. From these stones, they report the distribution of stone type by sex and age. I have made a graph from Table 2 of their publication.
Population Sex Ratio
The general population contains more males than females at younger ages (blue dots). By age 30-39 the two sexes are present in equal numbers. Thereafter, as men predecease women, their blue dots slump downward.
For all ages combined, the ratio of men to women is just under 1 (last blue dot at right).
Stone Former Sex Ratio
The blue bars show male to female ratios among stone formers. Remember this is counted from the sex of the person whose stone was analyzed. A survey based on symptomatic rates of stone passage, by contrast, might give different results altogether.
In childhood, men have slightly more stones than women (blue bar is above 1.0). In the teen years and up to age 39, women predominate over men (blue bars are below 1.0). After age 40 men predominate, until at age 90 and more, in this and perhaps most things, the sexes come into a near perfect alignment. Averaged over all of life, men have more stones, which appears to be because of their midlife excesses (Height of the ‘ALL AGES’ bar above 1.0).
The fraction of all stones formed (red dots; scale along the right axis) for both sexes combined is highest from age 20-69, with only a small fraction in childhood or old age.
Types of Stones
The men are on top, women on the bottom of the picture to the left.
Stones were classified using the system I have used on this site. Uric acid in any amount meant the stones were classified as uric acid stones, and likewise for any struvite or cystine.
CaOx stones preponderate among both sexes over all ages, except in women between ages 20 – 39 stones were about half CaOx and HA. With age, HA stone frequency fell in both sexes, so that most men, and most older women (over 40) have CaOx stones.
Brushite stones, in both sexes, are very uncommon. You can see them as triangles along the bottom of both graphs.
Over age 50, uric acid stones become a significant concern in both sexes.
Struvite stones, which always arise from infection with bacteria that possess urease, are more common in women than men, a fact known for ages.
The Mystery of Brushite
Brushite stones are rare but should be rarer still. I have written a whole article on brushite because it is so important and yet so evanescent. It forms first of all crystals in human urine. If pH is not too high, oxalate steals away its calcium atoms so it vanishes. If pH is high, HA does the same, and brushite vanishes.
Why, then, are there any brushite stone formers?
I do not know nor does anyone I know of. It is an open question that seems obscure but whose answer might well lead to some new understanding of how stones form.
The Importance of Brushite
Being the first crystal to form, brushite supersaturation is crucial for stone prevention, a fact not intuitive but worthy of special emphasis. Rare in stones, vanishing in most urine, yet brushite supersaturation is foremost in importance for clinicians and patients. The goal is a supersaturation below 1, so brushite cannot form. For those who want to know more about why, please look at the parent article.
Time and Shock Wave Lithotripsy
We (left hand figure below) and others have noted an increasing percent of CaP in stones over the past 30 years. In women (black dots) CaP percent is always higher than in men, but it has risen in both. For those of a quantitative bent, the time trend of stone CaP tested by ANOVA with post hoc contrasts was significant for both sexes, and women were higher than men throughout.
The number of shock wave procedures per patient adjusted for the number of stones and the years of stone disease rose with the percent of CaP in stones (Panel A of the figure below to the right) and the percent of CaP likewise adjusted for number of stones and duration of stones and sex rose progressively with the number of shock wave procedures (Panel B of figure to the lower right).
Not shown here, but of interest, the number of shock wave treatments was higher among BRSF than HASF suggesting a link between shock wave treatment and brushite stones.
One might infer from this set of graphs that the advent of shock wave lithotripsy caused the increase in phosphate stones, and there is nothing to contradict the idea. In fact, the very physiology of phosphate stone formation and the effects of shock waves on kidney function strongly support that idea as I shall show you.
Mechanisms of Phosphate Stone Formation
High Urine pH
As expected, percent CaP in stones (upper left panel of the figure below) rises with CaP SS. I have shown elsewhere on this site that stone crystals parallel urine supersaturations.
Because CaP SS depends powerfully on urine pH one expects and finds (upper middle panel) that urine pH tracks very closely with stone CaP percent. Urine calcium, volume, phosphate, and citrate excretions (remaining panels) had no important relationship to stone CaP percent.
But take a look at the urine calcium excretions (Upper right panel). They are very high on average. This is because a high fraction of all calcium stone formers have genetic (idiopathic) hypercalciuria. Risk for stone forming begins at a urine calcium of 200 mg/d in both sexes.
So you can think of CaP stones as a two hit model.
Genetic hypercalciuria promotes calcium stones, and urine pH controls the fraction of phosphate in the stones. High CaP SS and CaP stones require a urine pH significantly above 6 as shown in the upper middle panel.
Plaque and Plugs
Idiopathic CaP stone formers, and patients with stones from bowel disease, ileostomy, renal tubular acidosis, and primary hyperparathyroidism, form stones on plaque but also on plugs of HA that fill and damage the last millimeter or so of the nephron, the inner medullary collecting ducts and ducts of Bellini.
Although we are not certain, I think it is fair to say that the plugging of CaP stone formers is because more CaP crystals form in urine and can produce plugging. In a recent article I trace out how calcium phosphate actually forms, how it is a rapid process compared to calcium oxalate, and therefore more able to make plugs during the short times it takes for urine to pass out of tubules into the renal pelvis.
Distinctions Among the Three Idiopathic Calcium Stone Formers
We have published selected laboratory and tissue findings of CaOx, brushite and HA stone formers, in an attempt to clarify differences in how stones form, and amounts of tissue injury.
Numbers are small because each patient had been studied with intra-operative imaging of the renal papillae and papillary biopsy: 11 CaP (HASF), 25 BR (BRSF), and 30 CaOx (ICSF) stone formers.
As expected, urine pH was higher in the BRSF and HASF than in the ICSF, as was supersaturation (SS) for CaP. Incidentally, urine calcium (Ca) was also higher in both CaP groups than in the ICSF.
Mainly CaOx and BR stone formers formed plaque, and mainly CaOx SF form stones on it. About 8% and 6% of papillary surfaces were covered with plaque in ICSF and BRSF but only 0.8% among the HASF. CaOx stone formers had an average of 10 stones/patient attached to plaque, vs. only 3 plaque stones in 25 BRSF and 6 in 11 HASF stone patients: 10/ CaOx patient vs 0.12/brushite patient and 0.54/HA patient. These are 80 and 18 fold differences, respectively!
Plugging (‘deposits’ in the table) was absent in ICSF, but common in BRSF and HASF. Plug size averaged 1.6 mm2 in BSRF but only 0.4 mm2 in HASF – a 4 fold difference. The number of plugs was 3 times higher in HA vs. Br patients: 12 vs. 3/mm3 of tissue volume. BRSF formed fewer but much bigger tubule plugs.
Calyx dilation (a abnormal finding) estimated during surgery was higher in HASF than in the other two groups, and papillary injury (papillae are the parts of kidneys inside calyces) higher in both phosphate groups than in ICSF.
In the kidney cortex, far from where stones form, many CaP stone formers had scarring (TIF, tubular interstitial fibrosis) vs. very few CaOx patients. Brushite patients had most cortical damage.
So phosphate stone formers have injury involving the papillae and cortex, whereas CaOx stone formers have almost none.
Why is Urine pH High?
I wrote a whole article on how women raise their urine pH. They do it by absorbing from their food a higher fraction of its alkali content. No sense copying all that here, it is better to read the article. High GI alkali absorption is not easy to treat. Those alkali are nutrient – anions that cells metabolize to get energy.
We used a massive database of kidney stone formers to ask what happened to urine pH in men and women with age. The answer is it falls, in both sexes (women are circles, men triangles).
Why is a long story. We could exclude gain in BMI, loss of kidney function, and GI alkali as reasons, but could not find the reason itself. In fact, GI anion absorption rose with age, as if to offset the falling pH.
Here the important fact is on the graph – higher pH in women and in youth are an obvious cause of more CaP stones.
Shock Wave Lithotripsy (SWL)
No practical experiments permit us to measure effects of SWL on urine pH in humans.
For these reasons we turned to an animal model: The farm pig whose kidney is much like that of a human, and likewise is similar in size.
In these animals we could shock one kidney, and then compare the treated to control side at time intervals after the treatment, the untreated side being a perfect control as both kidneys are bathed by the same blood.
SWL Raised Urine pH.
Urine pH from the treated kidneys was 0.18 pH units higher than the control, meaning SWL had increased urine pH (first line of table under ‘Basal’).
SWL Damaged Kidney Tubule Function
There was a lot more.
Urine flow, and excretion of bicarbonate, potassium, chloride, sulfate, calcium, magnesium, sodium and oxalate all were higher from the treated side (bolded). This means that shock wave treatment affects tubule handling of multiple molecules, presumably because of injury.
Bicarbonate Losing Raised the pH
The higher urine pH could have been due to damage of final urine acidification in the collecting ducts or to high delivery of bicarbonate from higher up in the nephron so that acid secreted lower down was neutralized by a flood of bicarbonate.
To tell these apart we gave the pigs an acid load that lowered their blood bicarbonate and therefore filtration and downstream delivery (‘Acid load’ columns). Acid load brought urine pH and almost all other measurements to equality between the shocked and control kidneys (loss of bolding).
The tissues from the pigs showed widespread injury to the thick ascending limbs, and you can read the paper for details.
SWL Can Raise Urine pH by Damaging Kidney Tubules
The meaning of the work is clear.
After shock wave treatments the treated kidney may excrete excess calcium and produce a urine of higher pH than it would otherwise do. The effects are precisely those needed to produce calcium phosphate crystals. From the bladder urine, which mixes urine from both kidneys, one could never know this was happening.
It is possible that the advent of shockwave lithotripsy has contributed to the rise in CaP stones, and I hope that further science sorts out whether this hypothesis is false or true.
High Kidney Ammonia Production
Ammonia Production Regulates Urine pH
Kidneys excrete acid by making ammonia that can carry acid (protons) into urine without lowering urine pH. They also excrete acid by titrating urine phosphate, which does lower urine pH. If ammonia production goes down, from kidney disease, for example, urine pH has to fall so acid can be lost on phosphate.
Ammonia production relates itself to body acid load – from food and metabolism – so that the average urine pH is just around 6. But what would happen if regulation were abnormal so more ammonia than normal was made for a given bodily acid load?
Urine pH would rise.
CaP Stone Formers Make More Ammonia
The graph shows urine ammonia excretion from normals, and CaOx and CaP stone formers studied eating the exact same diet in a research center.
Fasting, all three groups are the same (left panels). Food increased urine ammonia in male CaP patients (#). Fed, the female CaP stone formers produce more ammonia than female normals (*, top right panel). So do the female CaOx stone formers. Ammonia production is governed by body acid load, which we measure as GI anion and urine sulfate – a residue of metabolic acid production. When we adjust ammonia for acid load (lower right panel) CaP stone male and female stone formers remain high compared to same sex normals.
We suspect the high urine pH that causes CaP stones arises in part from high ammonia production, perhaps an inherited trait.
Low Urine Citrate
studied men and women we could document a uniquely low urine citrate of CaP stone formers vs. normal people.
Low Citrate in CaP Stone Formers
Food increased urine citrate is normal women and all three male groups (#). With food, CaP stone formers of both sexes have urine citrate excretions below their same sex normal counterparts (*, upper right panel) as did female CaOx stone formers.
As is well known, citrate is lower in normal men than women (compare black bars; we did not choose to compare the sexes statistically).
Adjusted for GI alkali and urine sulfate, (lower right panel) low citrate is concentrated among male CaP and female CaOx stone formers. Normal men remain below normal women.
Male CaOx stone formers have abnormally high urine citrate with and without adjustment for systemic acid balance.
Abnormal Kidney Cell Citrate Handling
Alkali loads, most famously potassium citrate, raise urine citrate and is an established stone prevention. Citrate also raises urine pH, because the alkali appears in urine as bicarbonate. That is why potassium citrate is not an ideal treatment against CaP stones, and why we have for decades needed a controlled trial to see if it works or makes things worse.
But here we have a high urine pH coupled with low urine citrate, in male CaP and female CaOx and CaP stone formers. That points to something wrong with kidney cell regulation.
We measured serum citrate and glomerular filtration so we could calculate the fraction of filtered citrate excreted (FE Citrate), shown in the upper right panel of the graph at left.
FE citrate is low in female CaOx and CaP stone formers and in males with CaP stones. This means that CaP stone formers are reabsorbing abnormal amounts of citrate back from the filtrate. It is used by kidney cells to produce metabolic energy.
Adjusting for GI alkali absorption (lower right panel) removes the female abnormalities but makes the male one even more prominent.
That male CaOx stone formers have abnormally high urine citrate excretion with normal FE citrate is because their serum citrate concentration is higher, a fact for which we had no explanation.
CaP Stone Formers Have Proximal Tubule Abnormalities
Citrate reabsorption and ammonia production are linked in the proximal tubules of the kidneys as part of overall kidney regulation of bodily acid base balance. In general alkali loads raise urine pH and urine citrate, and reduce ammonia production, whereas acid loads do the opposite.
Here we have high pH and high ammonia production coupled with low urine citrate, more marked in male CaP patients but detectable among the women as well.
It is as though the cells perceive a need to produce more acid excretion (ammonia) and conserve potential alkali (citrate is metabolized to bicarbonate), but there is no need. So urine pH rises and converts calcium stones to their phosphate forms. The cause(s) of these proximal tubule abnormalities are not known.
Incomplete Distal Renal Tubular Acidosis (dRTA)
A Questionable Disorder
Some have proposed that CaP stone formers have high urine pH and low citrate as part of “Incomplete renal tubular acidosis”. In proof, when given extra acid they may not reduce urine pH as low as normal people. In my primary article on dRTA, I present contemporary evidence that acid loading creates a continuous spectrum of urine pH responses, even among normal controls, so it is not a good basis for diagnosis. It seems better to say that CaP stone formers have abnormal proximal tubule functions, and make those the focus of new science.
Heterozygotes of Familial dRTA
With one exception, hereditary dRTA arises from gene disorders of the main proton transporters or of carbonic anhydrase itself, and these disorders are in general recessive. They are recessive because you need two defective genes to knock out a transporter whereas one good gene copy will maintain function.
Of course dRTA causes massive CaP stones and kidney disease. But heterozygotes – meaning one good and one defective gene – from families with dRTA if studied in detail, may not lower urine pH normally. These might be diagnosed as ‘incomplete dRTA, because in fact that is what they are.
CaP Stone Formers are Not Like Incomplete dRTA
Unlike our CaP stone formers, urine ammonia is low in dRTA and heterozygotes from families of dRTA, when compared to their acid load – urine sulfate. Urine ammonia is never high. I suspect that some CaP stone formers have high urine pH because they are indeed heterozygotes of dRTA. Low ammonia may be a way to separate them from the high ammonia of routine CaP stone formers.
Risk of Conversion From CaOx to CaP Stones
Some patients gradually increase their stone CaP percent, often enough to alter their classification to CaP stone former. The opposite, conversion from CaP to CaOx stones must be very uncommon, as we have no cases to report. We wanted to know how to detect risk of conversion.
Who We Studied
From 4767 patients in our program, we collected all CaOx stone formers who had two or more stone analyses and clinical follow up data (445 patients). From these we selected all who had a last stone CaP% at least 20% higher than that of the first stone (62 patients). Men and women were combined because we had so few cases.
Of these 62 cases, 26 had had three initial (pre-treatment) 24 hour urine studies before they passed the stone whose CaP percent was at least 20% higher than their first stone. We labeled these transformers with prior laboratory work – labs before they transformed – as ‘TP’.
For controls we chose 181 patients whose first stones were >90% CaOx and who increased their stone CaP percent <20% between the first and last stone.
This figure shows the 26 TP cases and the 181 controls.
CaP% Was High at the beginning
Even though their initial stone CaOx percent was >50%, the 26 TP cases (black circles, upper left panel) had an average stone CaP of 10% before treatment, whereas it was much lower in the controls – who never added significant CaP.
During follow-up (upper left and middle panels) the 26 TP (black circles) increased their stone CaP markedly (average 10% to 79%, top left). The controls (gray triangles) hardly changed (-0.6% for controls, 69% change, for TP, upper middle panel).
Higher Urine pH Increased CaP SS
Urine pH and CaP SS before treatment and before conversion (upper right panel and lower left panels) and during treatment (lower middle and lower right panels) were higher in TP (black circles) than controls. CaP SS rose because we used potassium citrate as part of our treatment program.
SWL May Have Played a Role
ESWL associated with conversion: 112 of the 136 total cases with no ESWL procedures were controls, whereas only 21/41 cases with >2 ESWL were controls (X2=17, p<0.001). Furthermore, a predominance of ESWL procedures preceded the final stone (not shown here but shown in the paper), meaning ESWL could have been a causal factor.
Who is at Risk?
When stone CaP is above 10%, average 24 hour pH is as high as 6.3, or CaP supersaturation is above 2 before treatment risk of increasing stone CaP may be high. More than 2 ESWL procedures likewise. Given these risk factors in a CaOx SF perhaps one is prudent to treat as if CaP stones were already forming, so as to possibly prevent further stone CaP accumulation.
Prevention of Calcium Phosphate Stones
The objective is to lower CaP SS – reported with respect to brushite – below 1.
The main modifiable factors are urine volume, and calcium and citrate excretion. Because we cannot lower urine pH, the most crucial factor, we have to use what is left to achieve our goal. Likewise, because citrate regulation is abnormal in CaP stone formers, use of potassium citrate may not raise urine citrate so much as it raises urine pH, and therefore this otherwise valuable treatment can be ineffective.
Relative calcium stone risk falls to 1 (no excess risk) at about 2.3 l/d of urine volume. Given the limitations of our treatments, I usually strive for 2.5 l/d spread out over the waking hours. This is an achievable goal if patients understand why it is important for their stone prevention.
Reduced Calcium Excretion
Genetic hypercalciuria is very common among calcium stone formers. If we understand that relative risk of stones rises above 1 at a urine calcium of 200 mg/d, both sexes, our goal is to reduce urine calcium to or below that point.
Reduced Diet Sodium
Multiple articles on this site detail the power of diet sodium to control urine calcium and bone calcium balance. The US diet recommendations for sodium are 100 mEq (2300 mg)/day as a tolerable upper limit, and 65 mEq (1500 mg)/day as ideal. These values concern blood pressure and bone rather than kidney stones. But if we achieve an ideal diet sodium it will lower urine calcium as well as defend blood pressure and bone mineral. So I have no reservations about promoting the ideal diet sodium, but also am prepared for compromise in this fast food dominated world.
Reduced Diet Sugar
As for diet sodium, I have written extensively about sugar as a factor that raises urine calcium, abruptly after the sugar load and with proven increase in supersaturations. Once again, US guidelines call for reducing sugar intake, and there is no benefit to anyone from eating refined sugar in any form. So I am shameless in my zeal to encourage patients to eat as little of it as possible.
Drugs of this class lower urine calcium about 80 to 100 mg/d below the level predicted by sodium intake. They act in part to increase proximal tubule calcium reabsorption. They are trial proven agents to reduce calcium stone recurrence. We have shown thiazide drugs lower urine pH, a possible benefit.
I have often argued to use diet as much as possible before adding thiazide to avoid drug side effects. But phosphate stones are not easy to prevent, so far as I have observed, and they damage kidney tissue. Moreover, we have no trials – none. These patients may have been in trials but are doomed to perpetual minority status unless specifically a focus.
So I am not shy about adding thiazide after perhaps only one to two efforts at diet control, should CaP supersaturation remain above 1.
Why NIH has yet to fund a calcium phosphate stone prevention trial escapes me. I cannot imagine how this has not been a priority.
This drug will lower urine calcium below the level predicted by diet sodium intake. It may raise urine citrate excretion. But It may also raise urine pH.
Being as it is therefore able to raise or lower CaP supersaturation, I do not so much avoid using it as view it with a cold eye.
If thiazide is not attractive to a given patient I will try citrate and watch the effect on CaP supersaturation. CaP supersaturation is the final resultant of whatever changes it induces in urine calcium, pH, and citrate. If it indeed lowers CaP supersaturation, I am prone to use it but with appropriate 24 hour urine followup and an inextinguishable skepticism.
Reduced Diet Oxalate
I am aware that calcium oxalate in stones matters, and that even high phosphate stones often contain that crystal. If urine oxalate is high enough to confer risk – above 25 mg/d in both sexes – I make appropriate diet recommendations.
But patients cannot do everything all at once, so I generally put most emphasis on the calcium phosphate side. The exception is when urine oxalate is quite high – above 40 mg/d, for me – whereupon I do what I can with diet.
The objective is to lower CaP supersaturation below 1 in the 24 hour urine, and that is what I aim to achieve.
If fluids are enough, so be it. If not I add more treatments more or less as in the paragraphs above. Lacking trials, this is the best we can do. I watch supersaturation for calcium oxalate as a secondary endpoint, and if it is high enough to promote risk – above 3 – I attempt to lower it by reducing diet oxalate.
Monitoring is crucial. What we try to do may not be done because patients cannot or will not do it, so we have to know when to try another approach.
Put another way, for stone prevention, especially calcium phosphate stones, deliberation is reality.
I wish to thank Dr John Asplin for his careful reading of this article and suggestions for improvement.