effect of potassium-magnesium citrate on upper gastrointestinal mucosa

Effect of potassium-magnesium citrate on upper gastrointestinalmucosa

G. B. GONZALEZ, C. Y. C. PAK, B. ADAMS-HUET, R. TAYLOR & L. E. BILHARTZ*

Center for Mineral Metabolism and Clinical Research and *Department of Internal Medicine, University of Texas

Southwestern Medical Center at Dallas, Dallas, Texas, USA

Accepted for publication 17 September 1997

INTRODUCTION

Potassium supplements, frequently prescribed for the

prevention or treatment of hypokalemia, have been

known to damage the mucosa of the upper gastroin-

testinal tract.1±15 Early preparations of potassium

chloride in an enteric-coated tablet were found to cause

small bowel ulceration, stenosis and even perforation at

a rate of 40±50/100 000 patient years and their

clinical use was subsequently abandoned.16, 17 Later

preparations, using a slow-release wax-matrix oral

potassium supplement, although much less likely to

cause small bowel perforation (< 1/100 000 patient

years), still caused frequent upper gastrointestinal tract

erosions (43%) or ulceration (11%).5

Potassium-magnesium citrate (K-Mag) is a new drug

under development and is formulated as wax-matrix

tablets, which should be very helpful in the manage-

ment of recurrent calcium nephrolithiasis, as well as in

the prevention of thiazide-induced hypokalemia and

hypomagnesemia.18±20 Thus far, there are no data on

the potential harmful effects of K-Mag on the upper

gastrointestinal tract.

Based on reports of severe lesions during KCl therapy

concomitant with delayed gastrointestinal transit,

McMahon et al.1 introduced in 1982 the use of an

anticholinergic (glycopyrrolate) in the studies of gas-

trointestinal toxicity associated with KCl. The anticho-

linergic model is designed to mimic the effects of ageing,

diabetes, drugs and other disorders on the gastrointes-

tinal transit. These conditions, which form a common

SUMMARY

Background: Potassium supplements may cause mucosal

damage of the gastrointestinal tract.

Aim: To evaluate the effect of a new potassium

supplement, potassium-magnesium citrate (K-Mag), on

upper gastrointestinal mucosa and to compare it with

an older potassium supplement, potassium citrate

(Urocit-K).

Methods: A randomized and double-blind study was

conducted utilizing 36 healthy adults. Subjects were

randomized into three groups: K-Mag (70 mmol/day K,

35 mmol/day citrate and 17.6 mmol/day Mg); Urocit-K

(70 mmol/day K and 23.4 mmol/day citrate), and

placebo. All subjects took 5 tablets b.d. of the allocated

drug and 2 mg t.d.s. of glycopyrrolate for 7 days. On

day 8, stool was examined for occult blood, a symptom

score was calculated and an oesophagogastroduo-

denoscopy was performed. Mucosal lesions were scored

at ®ve anatomic sites.

Results: Demographic characteristics and symptom

score were similar in the three groups (< 10% with

more than mild symptoms). There were no signi®cant

differences in the endoscopic scores at any site examined

nor in the total scores among the three groups. Erosion

or ulcers were found in 18% of K-Mag, 23% of Urocit-K

and 17% of the placebo group.

Conclusion: Short-term use of K-Mag does not appear to

induce lesions in the upper gastrointestinal mucosa and

its oral tolerance is similar to Urocit-K or placebo.

Correspondence to: Dr C. Y. C. Pak, Center for Mineral Metabolism and

Clinical Research, University of Texas Southwestern Medical Center atDallas, 5323 Harry Hines Blvd., Dallas, TX 75235±8885, USA.

Aliment Pharmacol Ther 1998; 12: 105±110.

Ó 1998 Blackwell Science Ltd 105

clinical setting for the use of potassium supplements,

may enhance the toxicity of KCl in the upper gastro-

intestinal tract by increasing the time of exposure and

concentration of KCl in the stomach.7 Although it is

questionable whether this model is analogous to those

clinical conditions, the FDA has recommended its use

for the testing of oral potassium preparations.

In a previous study we showed that the irritative

potential on the upper gastrointestinal mucosa of a wax-

matrix formulation of potassium citrate, in use since

1985, was similar to that of an analogue formulation of

potassium chloride (Slow K).14 The present study was

conducted to determine the effects of a new drug,

potassium-magnesium citrate, on upper gastrointestinal

mucosa and to compare it to the older potassium

supplement, potassium citrate, and a placebo.

MATERIALS AND METHODS

Subjects

Forty-two healthy adults with no history of chronic

gastrointestinal disorders and without the use of

antacids, antisecretory or non-steroidal anti-in¯amma-

tory drugs (NSAIDs) were screened using a SMA-20 and

a qualitative test for IgG in serum for Helicobacter pylori

(Smith Kline Beecham Clinical Laboratories, Dallas, TX).

Any subject using the above drugs in the 2 weeks before

the study was excluded. Only subjects with both tests

normal were eligible for this study.

Study protocol

The study was conducted as a placebo-controlled and

double-blind trial in an outpatient setting. Subjects were

randomized into three groups: Group 1, potassium-

magnesium citrate (K-Mag): 5 tablets b.d. (70 mmol/day

K, 35 mmol/day citrate and 17.6 mmol/day Mg); Group

2, potassium citrate (Urocit-K): 5 tablets b.d. (70 mmol/

day K and 23.4 mmol/day citrate); and Group 3,

placebo: 5 tablets b.d. (K-Mag excipient only). Tablets

in the three groups were identical in appearance and

were formulated as wax-matrix tablets. All subjects were

instructed to take the drug before breakfast and at

bedtime, at least 2 h after the last meal, for 7 days. In

addition they took glycopyrrolate 2 mg t.d.s. to delay

gastric emptying.1 Alcohol, tobacco and spicy food were

restricted during the study. NSAID use was not allowed.

On the morning of day 8 all volunteers had a faecal

occult blood test (FOBT) (Hemoccult) performed, a

diagnostic oesophagogastroduodenoscopy (EGD) and a

symptom score calculated on the basis of a diary. Drug

compliance was assessed through count of returned

tablets and by asking each subject whether the medica-

tion regimen had been followed.

Endoscopic evaluation

Using lidocaine spray as a topical anaesthetic to the

pharynx, an Olympus GIF 100 video endoscope was

inserted and advanced to the third portion of the

duodenum. Upon withdrawal, the mucosa was carefully

examined and photographed in ®ve anatomic areas of

the upper gastrointestinal tract: post-bulbar duodenum,

duodenal bulb, antrum, fundus/body of the stomach

and the oesophagus. Any mucosal lesions were noted

and scored using a modi®ed Lanza21 grading system

(Table 1). A submucosal haemorrhage was de®ned as a

red macular lesion of any size. An erosion was de®ned

as a white-based disruption of the mucosal surface with

no perceptible depth and an ulcer was de®ned as a

disruption of the mucosa with obvious depth.

Endoscopic scores were measured for each of the ®ve

anatomic areas and summed to give the total endo-

scopic score. All 36 endoscopies were performed and

scored by a single academic gastroenterologist (LEB)

with experience in more than 5000 upper endoscopies

who was blinded as to which drug the subjects had been

taking.

Symptom evaluation

Symptoms were assessed by means of a symptom diary

and subject interview (by a single-blinded investigator)

according to the following scale:21 0 � no symptoms;

1 � minimal symptoms on 1 day; 2 � minimal symp-

toms on more than 1 day or moderate symptoms on

1 day; 3 � moderate symptoms on more than 1 day or

Table 1. Endoscopic score

Grade Description

0 Normal gross appearance or erythaema only

1 One or more submucosal haemorrhages

2 Erosion � submucosal haemorrhage

(2 points for each erosion)

3 Ulcer � submucosal haemorrhage

(3 points for each ulcer)

106 G. B. GONZALEZ et al.

Ó 1998 Blackwell Science Ltd, Aliment Pharmacol Ther 12, 105±110

severe symptoms on 1 day; 4 � severe symptoms on

more than 1 day. `Severe' was de®ned as interfering

with activity and `minimal' as noticeable but otherwise

causing no concern.

STATISTICAL METHODS

Based on a previous study,14 it was estimated that 10±

12 volunteers would be required to detect a difference of

three or more points (considered to be clinically

relevant) between groups in the primary outcome

variable (total endoscopic score), assuming a standard

deviation of 2.5 and power of 0.80 at the 0.05

signi®cance level. Continuous variables, as well as the

total endoscopic score and clinical score were compared

between groups using the exact Kruskal±Wallis test

statistic. For variables where the result was zero for

most respondents (such as endoscopic scores at each site

and symptom evaluation), the scores were collapsed

into two or three categories and were compared among

the three groups with Fisher's exact test. Spearman

correlation coef®cients were computed to assess the

association between symptom and endoscopic scores.

Statistical analysis was performed using SAS version

6.12 (SAS Institute Inc., Cary, NC) and StatXact-Turbo

version 2.11 (Cytel Software Corporation, Cambridge,

MA).

Ethics committee

All subjects gave written informed consent before the

study, which was approved by the Institutional Review

Board of the University of Texas Southwestern Medical

Center at Dallas.

RESULTS

Among 42 screened subjects, ®ve were not eligible due

to a positive Helicobacter pylori test. One volunteer was

withdrawn on day 4 of the study due to pregnancy.

Thus, 36 subjects completed the protocol (mean age

28.9 years, range 22±56; 26/10 male/female ratio).

The three treatment groups were comparable in demo-

graphic characteristics as well as with respect to drug

compliance (Table 2).

Symptom evaluation and blood loss

The symptom score and side-effects were similar in the

three groups. No subject reported severe symptoms in

the K-Mag group and only one subject in each of the

other groups reported severe symptoms. Only one

subject, receiving placebo, had a positive faecal occult

blood test; her symptom score and endoscopy score were

zero (Table 3).

Endoscopic ®ndings

There were no statistically signi®cant differences in the

endoscopic score at any site examined (Table 4) nor in

the total score between the three treatment groups

Table 2. Demographic characteristics and

drug complianceGroup K-Mag Urocit-K Placebo

n 13 11 12

Age (years) 28.3 � 8.7 31.4 � 10.4 27.6 � 4.0

Gender (male/female) (%) 69/31 64/36 83/17

Race (Caucasians/others) (%) 77/23 64/36 75/25

Drug compliance* (%) 98.9 97.9 98.8

Glycopyrrolate compliance (%) 99.3 97.0 96.6

* � [(tablets prescribed ) tablets returned) ´ 100]/tablets prescribed.

Table 3. Clinical tolerance, symptoms referred and faecal occult

blood

Group K-Mag Urocit-K Placebo

n 13 11 12

Clinical score* (%)

0 62 36 58

1±2 38 55 34

³ 3 0 9 8

Symptoms referred (%)

Epigastric pain 23 27 8

Cramps 15 18 0

Loose stools 8 9 0

Heartburn 8 0 17

Nausea 0 9 8

Others 8 18 17

Faecal occult

blood positive (%)

0 0 8

* See de®nition in text.

POTASSIUM-MAGNESIUM CITRATE AND GASTRODUODENAL DAMAGE 107


(Figure 1). For any group, the most frequent ®nding

was a normal EGD. Trivial or no mucosal pathology

(total score of 0 or 1) was seen in 10 out of 13 (77%) in

the K-Mag group, nine out of 11 (82%) in the Urocit-K

group and nine out of 12 (75%) in the placebo group. A

shallow ulcer less than 3 mm in size was seen in the

post-bulbar duodenum of one subject (from the K-Mag

group) with a zero symptom score and a negative faecal

occult blood test. There were no ulcers in others

subjects. Erosions were present in three out of 13

subjects (23%) in the K-Mag group, two out of 11 (18%)

in the Urocit-K group and two out of 12 (17%) in the

placebo group. No signi®cant correlation was found

between symptoms and endoscopic score.

DISCUSSION

In this study, potassium-magnesium citrate given for

7 days at a dose of 70 mmol/day of K in wax-matrix

tablets did not appear to induce lesions in the upper

gastrointestinal mucosa; its oral tolerance was similar

to placebo or to potassium citrate at the same dose of

potassium.

Enteric-coated tablets, the ®rst generation of solid KCl

supplements, were withdrawn from the market in 1965

because of their gastrointestinal toxicity. Small bowel

ulceration with haemorrhage, perforation and strictures

were reported at a rate of 40±50/100 000 patient

years.16, 17 Extensive experimental and clinical investi-

gations identi®ed the release and absorption of high

concentrations of potassium over short segments of

small bowel acting upon submucosal and mural veins to

produce spasm or paralysis with a resultant haemor-

rhagic infarction as the mechanism of damage.22±24

The same group of investigators later reported, using

similar experimental models, a signi®cantly higher

safety for potassium citrate and gluconate tablets versus

KCl.25 A possible explanation for this is that organic

anions (citrate and gluconate) are more slowly absorbed

across the intestinal wall, due to their large molecular

weight, and would therefore hold potassium within the

intestinal lumen, diminishing the concentration of

potassium in the intestinal wall veins at any one time.25

The slow-release KCl formulations in current use are

composed of a wax matrix forming a porous mass, with

small compartments in which KCl is incorporated. During

passage through the gastrointestinal tract, KCl is slowly

released and absorbed.12 Although the slow-release KCl

formulations were considered safe and therefore widely

prescribed, McMahon in 1982 reported a 55% incidence

of lesions in the mucosa of the upper gastrointestinal tract

directly attributable to the administration of 96 mmol/

day of K as wax-matrix KCl tablets after 1 week in healthy

men. Concomitant administration of glycopyrrolate, to

delay gastric emptying and mimic conditions that

enhance the toxicity of KCl worsened the severity of

mucosal injury.1 However, this study did not include a

placebo group, therefore the true incidence of mucosal

lesions associated with KCl was not known.

Table 4. Endoscopic score by rank in every site examined

Group K-Mag Urocit-K Placebo

n 13 11 12

Score* rank 0 1±3 > 3 0 1±3 > 3 0 1±3 > 3

Percentage of subjects

Oesophagus 100 0 0 91 0 9 92 8 0

Fundus-body 92 8 0 91 9 0 75 25 0

Antrum 85 15 0 73 18 9 84 8 8

Duodenal bulb 92 8 0 91 9 0 100 0 0

Post-bulbar 84 8 8 100 0 0 92 8 0

* See de®nition in text.

Figure 1. Total endoscopy score.



Prompted by these ®ndings, several groups, using the

same model, have studied the effects of different

potassium doses and formulations in order to clarify

the safety of potassium supplements over the mucosa of

the upper gastrointestinal tract.2±15 Table 5 shows a

summary of those studies comprising only placebo-

controlled trials.3, 5, 7, 10, 14 Since erosion or ulcers are

probably the most relevant ®ndings in this kind of study

and are less dependent on the degree of subjectivity of

EGD, non-erosive changes (oedema, hyperemia or

submucosal haemorrhages) were not included to obtain

the corresponding incidence of lesions. Despite these

criteria, results range from a higher incidence of lesions

associated only with wax-matrix tablets5, 10, 14 or with

any solid potassium formulations,3 to no difference with

respect to placebo,7 as also observed in our study. Some

of these differences may be explained by variations in

potassium dose, study design or the skills of endo-

scopists. However, even for positive studies, the clinical

relevance of ®ndings may be questioned, since the

extent and number of lesions appeared to be minor.

In the above placebo-controlled studies of potassium

supplements and gastrointestinal toxicity, there was no

correlation between symptomatic complaints and cor-

responding endoscopic lesions. No one had occult

bleeding or withdrew due to complications during the

trials. All these ®ndings were also true for our study and

illustrate the subclinical character of such lesions.

Although these results were based on data from healthy

volunteers studied under controlled conditions and over

a short period, they agree with data from a surveillance

program for wax-matrix KCl supplements that found

only one subject with upper gastrointestinal bleeding

among 1050 KCl users.28

Our study has some limitations. It was conducted as an

outpatient study and therefore there was little control

over diet and administration of tablets. We did not

perform a baseline EGD, therefore some of the endo-

scopic ®ndings might be pre-existent. The ideal design

for this study should include a baseline endoscopy.

Thus, subjects with signi®cant lesions in the upper

gastrointestinal tract could be accurately excluded.

However, the lack of baseline endoscopy does not

invalidate our conclusions, since ours was a randomized

placebo-controlled study, and no differences were

observed between any group after the intervention.

Moreover, the percentage of subjects with lesions in the

placebo group was similar to other reports that included

a baseline endoscopy,5, 7 as Table 5 shows. It means

that our way of selecting subjects was useful in

excluding signi®cant pathology of the upper gastroin-

testinal tract. Although our sample size was small, the

study was designed with enough power to detect a

clinically relevant difference. In fact, as was mentioned,

in some of previous studies, signi®cant differences were

detected using a similar sample size.

In conclusion, short-term use of potassium-magne-

sium citrate does not appear to induce lesions in the

upper gastrointestinal mucosa and its oral tolerance is

similar to potassium citrate or placebo.

Table 5. Incidence of upper gastrointestinal mucosal erosion or ulcers in placebo-controlled studies of potassium supplements

Paterson3 MacMahon5 Alson7 Sinar10 Pak14 Gonzalez*

Volunteers by group 15 15 15 30 10 11±13

Dose of K+ by formulation 96 mmol/day 24 mmol/day 72 mmol/day 60 mmol/day 96 mmol/day 70 mmol/day

of K+ ´ 1 week of K+ ´ 1 week of K+ ´ 1 week of K+ ´ 2 week of K+ ´ 1 week of K+ ´ 1 week

% lesions % lesions % lesions % lesions % lesions % lesions

Formulation

KCl wax-matrix 40 33 20 50 50

K citrate wax-matrix 40 18

K-Mg citrate wax-matrix 23

KCl microencapsulated I 13 7 13 7

KCl microencapsulated II 7

KCl porous-membrane 27

KCl extended-release cap. 13

KCl extended-release tab. 20

KCl liquid 0 0 13 20

Placebo 0 20 27 3 0 17

* Present study.

POTASSIUM-MAGNESIUM CITRATE AND GASTRODUODENAL DAMAGE 109


ACKNOWLEDGEMENTS

This study was supported by United States Public Health

Service Grants MO1-RR00633 and PO1-DK20543, and

Mission Pharmacal Co. Dr Gonzalez is a post-doctoral

research fellow supported in part by funds from the

School of Medicine, Catholic University of Chile and

Hospital Dr SoÂtero del RõÂo, Santiago, Chile. The authors

wish to thank Mr John Poindexter for his assistance in

the preparation of the manuscript.

REFERENCES

1 McMahon FG, Akdamar K, Ryan JR, Ertan A. Upper gastro-

intestinal lesions after potassium chloride supplements: a

controlled clinical trial. Lancet 1982; i: 1059±61.

2 Barkin JS, Harary AM, Shamblen CE, Lasseter K. Potassium-

chloride and gastrointestinal injury. Ann Intern Med 1983;

98: 261±2.

3 Patterson DJ, Weinstein GS, Jeffries GH. Endoscopic compari-

son of solid and liquid potassium chloride supplements. Lancet

1983; ii: 1077±8.

4 Ryan JR, McMahon FG, Akdamar K, Ertan A, Agrawal N.

Mucosal irritant potential of a potassium-sparing diuretic and

of a wax-matrix potassium chloride. Clin Pharmacol Ther

1984; 35: 90±3.

5 McMahon FG, Ryan JR, Akdamar K, Ertan A. Effect of po-

tassium chloride supplements on upper gastrointestinal mu-

cosa. Clin Pharmacol Ther 1984; 35: 852±5.

6 McLoughlin JC. A study on the effects of a combined frusemide

and controlled release potassium tablet (Diumide-K ContinusÒ

tablets) on gastrointestinal mucosa. Br J Clin Pract 1984; 38:

99±101.

7 Alsop WR, Moore JG, Rollins DE, Tolman KG. The effects of

®ve potassium chloride preparations on the upper gastroin-

testinal mucosa in healthy subjects receiving glycopyrrolate.

J Clin Pharmacol 1984; 24: 235±9.

8 Kendall C, Krantz KD, Berger A, Sharp T. Endoscopic evalu-

ation of slow-release potassium chloride preparations. Clin

Pharmacol Ther 1985; 38: 28±30.

9 McLoughlin JC. Effects on upper gastrointestinal mucosa of

three delivery systems of potassium as supplement to fruse-

mide administration. J R Soc Med 1985; 78: 459±62.

10 Sinar DR, Bozymski EM, Blackshear JL. Effects of oral potas-

sium supplements on upper gastrointestinal mucosa: multi-

center clinical comparison of three formulations and placebo.

Clin Ther 1986; 8: 157±63.

11 Moore JG, Alsop WR, Freston JW, Tolman KG. The effect of

oral potassium chloride on upper gastrointestinal mucosa in

healthy subjects: healing of lesions despite continuing treat-

ment. Gastrointest Endosc 1986; 32: 210±12.

12 Haxholdt H, Hey H, Lech Y, Matzen P. Endoscopic comparison

of the effect of two formulations of potassium chloride and

placebo on gastric mucosa. Curr Ther Res 1986; 40: 561±6.

13 Lech Y, Hey H, Jorgensen F, Matzen P, Ostergaard O. Evalu-

ation of the ulcerogenic effect of potassium chloride by en-

doscopy and fecal blood loss. J Clin Pharmacol 1987; 27:

206±9.

14 Pak CYC. Citrate and renal calculi. Min Electrolyte Metab

1987; 13: 257±66.

15 McLoughlin JC. Comparative effects of three delivery systems

of oral potassium on upper gastrointestinal mucosa. Br J Clin

Pract 1987; 41: 865±8.

16 Hutcheon DE. Bene®t±risk factors associated with supple-

mental potassium therapy (Editorial). J Clin Pharmacol 1976;

16: 85±7.

17 Trechot P, Netter P, Moore N, Vidrequin A, Boissel P, Voltz C.

Reasons for the removal of an enteric-coated slow release

potassium preparation from the market in France (Letter). Am

J Gastroenterol 1990; 85: 330±1.

18 Pak CYC. Citrate and renal calculi: an update. Min Electrolyte

Metab 1994; 20: 371±7.

19 Pak CYC, Koenig K, Khan R, Haynes S, Padalino P. Physico-

chemical action of potassium-magnesium citrate in nephrol-

ithiasis. J Bone Miner Res 1992; 7: 283±7.

20 Ettinger B, Pak CYC, Citron JT, VanGessel A. Randomized trial

of potassium magnesium citrate in the prevention of recurrent

calcium oxalate nephrolithiasis. VIII International Sympo-

sium on Urolithiasis 1996 (Abstract Book), pp. 437±9.

21 Lanza FL, Graham DY, Davis RE, Rack MF. Endoscopic com-

parison of cimetidine and sucralfate for prevention of napr-

oxen-induced acute gastroduodenal injury. Dig Dis Sci 1990;

35: 1494±9.

22 Boley SJ, Schultz L, Krieger H, Schwartz S, Elguezabal A, Allen

AC. Experimental evaluation of thiazides and potassium as a

cause of small-bowel ulcer. J Am Med Assoc 1965; 192:

763±8.

23 Boley SJ, Allen AC, Schultz L, Schwartz S. Potassium-induced

lesions of the small bowel: I. Clinical aspects. J Am Med Assoc

1965; 193: 997±1000.

24 Allen AC, Boley SJ, Schultz L, Schwartz S. Potassium-induced

lesions of the small bowel: II. Pathology and pathogenesis.

J Am Med Assoc 1965; 193: 1001±6.

25 Boley SJ, Schultz L, Schwartz S, Katz A, Allen AC. Potassium

citrate and potassium gluconate vs. potassium chloride. Ex-

perimental evaluation of relative intestinal toxicity. J Am Med

Assoc 1967; 199: 215±17.

26 Aselton PJ, Jick H. Short-term follow-up study of wax matrix

potassium chloride in relation to gastrointestinal bleeding.

Lancet 1983; 1: 184.



effect of potassium-magnesium citrate on upper gastrointestinal mucosa

Documents