ffets of Hemodialysis and of Gluose-Insulin Administration on Plasma Potassium and on the letroardiogram By Borys Surawiz, M.D., Arthur S. Kunin, M.D., and than A. H. Sims, M.D. With the tehnial assistane of Y. K. Starheska xtraellular potassium (K) onentration may be lowered by removing K from the body or by transferring K from the extraellular to the intraellular ompartment. Hemodialysis against a K-free solution removes K from the body 1 ' 2 whereas the intravenous administration of gluose with insulin results in a shift of extraellular K into the ells. 3 Both methods for reduing extraellular K have been widely applied to the linial treatment of hyperpotassemia, but no data exist to show the relative effetiveness of eah proedure on the eletrial ativity of the heart. Results obtained in the present study show that eletroardiographie hanges following hemodialysis are essentially the same as those brought about by omparable redution of plasma K onentration aused by infusion of gluose and insulin. Methods Female mongrel dogs weighing 10 to 12 kg, anesthetized with halothane and ventilated with 100% oxygen by means of a yli respirator, were subjeted to four proedures, arried out in random order at weekly intervals: 1. Hemodialysis with a single oil of a Kolfi-type twin oil hemodialyzer (Travenol) against a K-f ree bath for four hours. The dialyzing fluid had the following omposition: Na 133, Cl 103.5, Ca 5 and Mg 1.5 niq/l, gluose 400 mg/100 ml, at a ph of 7.4. 2. Intravenous infusion of 10% gluose with 0.1 unit of rystalline insulin per ml From the Division of xperimental Mediine and the Metaboli Unit of the Department of Mediine, University of Vermont, College of Mediine, Burlington, Vermont. Supported by Grants H-4858 and A-4340 from the U. S. Publi Health Servie, Training Grant 2A-2086, and grants-in-aid from the Amerian Heart Assoiation and the Vermont Heart Assoiation. Reeived for publiation July 20, 1962. Cirulation Researh. Volume Xlf, February 10GS at a rate of 5 ml/min for four hours. 3. Hemodialysis for-two hours, followed by gluose-insulin infusion for two hours. 4. Gluose-insulin infusion for two hours followed by hemodialysis for two hours. Two dogs were subjeted to all four proedures, one dog to three proedures, three dogs to two proedures, and one dog to one proedure. The dogs were allowed to eat ad lib between experiments. Their general ondition at the beginning of eah experiment was onsidered to be good, and the initial plasma K was within normal limits. Blood ph during eah proedure was maintained within normal limits. The weights were monitored throughout the proedures and kept onstant by adjusting the rate of infusion of fluid and the filtration pressure of the oil. Blood pressure was likewise monitored, and abnormal hypotension was avoided by initiating dialysis with a oil pressure equal to arterial pressure. The onentrations of K and Ca in plasma, and K in the bath and urine and the urinary volumes were determined hourly. K onentrations were determined by internal standard flame photometry, Ca was determined by DTA titration. 4 The standard deviation of repliate determinations was ± 0.06 mq/l for K and ± 0.05 mq/l for Ca. Heparin was employed as the antioagulant. There was no orretion of eletrolyte onentration for protein ontent. Limb leads and lead V3 of the eletroardiogram were reorded with a diret-writing eletroardiograph (Cambridge Versa-seribe) at a paper speed of 50 mm/see. CG intervals were measured to the nearest 0.01 se and 0.5 mm on ten-fold enlargements of the original reords. The preordial eletrode was maintained in a onstant position throughout the experiment. The measurements of heart rate, duration of QBS omplex, P-R interval, Q-T interval and amplitude of the P wave were made in leads II and V3. Results are given as the average of these values in two leads. Analysis of CG reords was made without knowledge of results of the hemial studies. The duration of QRS was measured as desribed previously. 5 There was no diffiulty with measurement of the QT intei-val as no U waves were present. 145
146 STJRAW1CZ, KUNIN, SIMS -~ and.3 S lure! p? & i i o -1 I O 9 Si 0.01 st J C w = W C S.D. S Q 1 a w mea D Vi ean o 03 C S s 01 s s d S o 8 + I 1O i-l ri O r-l O I + + + The means, standard deviations, single and multiple orrelation oeffiients of the data were alulated with the aid of a digital omputer, IBM 1620. The "t" test was applied to evaluate the signifiane of the differenes between two sample means and the signifiane of orrelation oeffiients (). P values less than 0.05 were onsidered signifiant. Results PLASMA POTASSIUM Results are presented in detail in table 1. The mean absolute K onentrations at the end (K p end) of two hours of gluose-insulin infusion and at the end of two hours of dialysis were nearly idential. The lowest mean K,, onentration was present after two hours of dialysis following two hours of gluose-insulin infusion. The differene between this value and all other mean K,, onentrations was statistially signifiant. The mean hange in K p onentrations (K pa ) at the end of two hours of gluoseinsulin infusion and at the end of two hours of dialysis were of similar magnitude. The mean K p onentration did not hange signifiantly during the seond two hours of gluoseinsulin infusion, seond two hours of dialysis, or during the two-hour gluose-insulin infusion following dialysis. However, there was derease in mean K p onentration when dialysis followed the gluose-insulin infusion. The dogs were anuri during the dialyses; K losses were thus restrited to diffusion from blood to dialyzing solution. Removal of K proeeded at a uniform rate of 6.6 to 7.9 mq/hr, resulting in ellular K losses. During infusion of gluose-insulin the urinary K losses were smaller than the assumed derease in the total extraellular K whih suggested a slight ellular uptake of K during the first two hours of infusion. During the seond two hours slight ellular losses probably ourred. There was a slight but signifiant derease of the mean hange in plasma Ca onentration (Ca p4 ) after gluose-insulin infusion. o a 2 a. ^>'HJ& ^ o 'i "i tic C ^ ^S ^: ^ Kp o e> q < s - * * * * g MM"." M CO aj -j i q ~.S LCTROCARDIOGRAPHIC CHANGS The most obvious BCG hanges observed during development of hypopotassemia onsisted of an inrease in amplitude of the P Cirulation Researh, Volume XII, February 196S
POTASSIUM AND LCTROCARDIOGRAM 147 wave, widening of the QRS omplex and lengthening of the Q-T interval. The latter resulted from inreased duration of the terminal part of the T wave. In ertain instanes the T wave beame diphasi or nothed, and, in analogy to the human pattern of hypopotassemia one ould onsider the first hump as T wave and the seond as U wave. However, suh hanges were not onstant, and the sequene of development learly suggested gradual inrease in the duration of the T wave. Moreover, no U wave was present in the ontrol traings. Therefore, we made no attempt to exlude the terminal T-wave hump from measurements of the Q-T interval. Prolongation of the P-R interval, depression of the S-T segment and various heart rate hanges were also frequently observed. There were no eetopi beats or seond degree A-V bloks. All of the observed CG hanges were attributed to hypopotassemia beause they were abolished within 5-.1.5 minutes by intravenous administration of potassium hloride. This was arried out after the ompleted proedure, and it was noted that the rapidity with whih the ontrol pattern returned was related to the speed of KC1 infusion rather than to the administered amount of potassium. The signifiane of the CG hanges was evaluated in four different ways. 1. Correlations Betiveen Control Values. The orrelation oeffiient (R) between mean QRS 4 and mean QT,,^ was 0.51 and between mean QRS a and mean P A 0.52. There was no orrelation between hanges in QRS, QT C and P amplitude with hanges in the mean P-R A or mean R-R i. 2. Correlation Between Mean CG Changes During Various Proedures. The mean QRS A was signifiantly greater after the first two hours of gluose-insulin infusion and first two hours of dialysis than after all other proedures. The mean QT i was signifiantly greater after the first two hours of gluoseinsulin infusion and after the first two hours of dialysis than after the seond two hours of gluose-insulin infusion, or after two hours of dialysis following gluose-insulin infusion. Cirulation Researh, Volume XII, February 1963 The mean P a was signifiantly, greater after the first two hours of gluose-insulin infusion and the first two hours of dialysis than after the two-hour gluose-insulin infusion following dialysis. 3. Correlation Betiveen Mean GG Changes and Mean Potassium Changes. The CG hanges showed moderate (R = 0.4-0.6) to weak (R = 0.3-0.4) orrelation with mean K A but no orrelation with mean K p end. Correlation between mean QRS^ and mean K i (R = 0.54) was somewhat stronger than orrelation between mean P a, mean K p4 (R -0.48), mean QT i and mean K pi (R = -0.36). 4. Correlation Between Small and Large CG Changes vs. Small and Large K Changes. All 36 values of a given CG hange and K onentration hange were divided into two groups. The group with values less than the mean was designated as small, and the group with values greater than the mean as large. The groups with small and large K p4 differed signifiantly in their mean QRS A but not in mean QT i and P A. Conversely, groups with small and large QRSi differed signifiantly in their K pi. The same was true of the groups with small and large QT i. The results of these different methods used for orrelation of the hanges in the CG with hanges in K may be summarized as ; follows: during lowering of K,, onentrations, the CG showed a fairly onsistent inrease in duration of the QRS omplex and' the QT C interval and an inrease in amplitude of the P wave while hanges in heart rate' and P-R interval had no onsistent diretion. The CG hanges showed a relation to the derease in K p but not to absolute K p onentrations. The inrease in QRS duration provided the most sensitive index of K p _j while lengthening of QT C and inrease in P amplitude were less sensitive. Disussion The animals studied had normal plasma K onentrations and presumably had normal stores of intraellular K; the results of our study may therefore not be diretly applia-
148 SUAWICZ, KUNIN, SIMS 5-4-61 CONTROL -r m Duursts 5-18-61 CONTROL i h n GLUCOS - I - - - - QRS A 10 QT A +150% PA +0.5 PR A - 0.5 RR A - 4.0 Ks ontrol = 3.7mq/L Ks enrt = 2 4mq/L TotalBody KA- -360mq QRS A 1.0 QT A +16.0% PA 0.5 PR A RR A Ks ontrol Ks end Total Booy K 0.0 8.0 2.4mq/L 0.0 FIGUR 1 letroardiogra/phi hanges after four hours of dialysis and troo hours of gluoseinsulin infusion are nearly idential in the same dog. QRS^ = differene in the duration of the QRS omplex in 0.01 se; QT C$ = differene in the orreted QT duration in per ent. P A = differene in F wave amplitude in millivolts. P-R± = differene in the duration of the P-R interval in 0.01 se; RR± = differene in the duration of the R-R interval in 0.01 se; K s ontrol = ontrol onentration of plasma K; K s end = final onentration of plasma K; Total body if differene in the total body K ontent. ble to hyperpotassemi and potassium-loaded animal and man. Gluose-insulin infusion and hemodialysis had the same effet on plasma K onentration; both proedures aused an equal derease during the first two hours, and no hanges during the seond two hours. It has been observed * " 9 that hemodialysis has a diphasi effet on body potassium. During the first phase there is a rapid derease in extraellular K onentration while, during the seond, the ellular depletion ontinues but extraellular K onentration remains onstant at a new level. In the study of Weller and assoiates 7 this new level was 50% of normal or about 2.0 mq/l. In the study of Young and Daugherty 9 the rate of potassium removal was somewhat more rapid and the new level was about 1.6 mq/l. We have observed the same sequene of events but in our ases this new level was about 2.5 mq/l, possibly beause our rate of potassium removal was somewhat slower than that of Weller and o-workers. We found that the same new level of K p onentration was also attained by gluose-insulin administration, although the mehanism of K p derease was different from that in dialysis. Gluose-insulin infusion aused no plasma K lowering after the latter was dereased by dialysis. On the ontrary, plasma K onentration inreased slightly during this proedure as it did during the seond two hours of the infusion alone. The only proedure whih further dereased the plasma K onentration below the new level was dialysis after gluoseinsulin infusion. The mean K p after this proedure was 1.8 mq/l. This observation suggests that suh a ombined proedure is most effetive for rapidly lowering plasma K onentration. Cirulation Volume Xtl, February J963
POTASSIUM AND LCTROCARDIOGRAM 149 Within the limits of the observed potassium shifts, hanges in intraellular K had no effet on the eletroardiogram. However, the magnitude of assumed intraeelhilar inreases and dereases in the aute experiments was very small in relation to the large amount of intraellular potassium. The CG- hanges during hypopotassemia were the same whether the dog lost or gained potassium (fig. 1). "We do not know the extent of the myoardial partiipation in these ellular losses and gains. However, an inferene an be made from studies of other investigations. Habib et al. 10 found about 7% inrease of myoardial K ontent during hemodialysis in dogs with hypopotassemia. and total body K losses of the same magnitude as in our dogs. Tu et al. also found an inrease in myoardial potassium during dialysis but this inrease was desribed as small and insignifiant. 8 Rebare et al. 11 found inrease of myoeardial potassium of the order of 17% in the left ventrile and 20% in the right ventrile after administration of gluose with insulin. Comparing the dosages used by these investigators with our dosages, we find that per hour the dogs of Rebare et al. reeived 16 g of gluose and 100 units of insulin and our dogs 30 g of gluose and 30 units of insulin. It seems unlikely that hanges in myoardial K ontent of the magnitude quoted in the above studies would appreiably affet the CG. We onsider that in aute hypopotassemia, produed experimentally by hemodialysis or gluose-insulin infusion, the hanges in potassium gradient aross the myoardial fiber are largely determined by hanges in extraellular K onentration whih were of the order of 50% or more. Therefore, it is not surprising that the only K hange refleted in the BCG was the hange in extraellular potassium onentration. Our experiments an be likened to an aute hypopotassemia whih develops in patients during treatment of diabeti aidosis or an attak of familial periodi paralysis. In these situations the BCG reflets hanges in plasma K onentration. 12-13 The CG pattern of hypopotassemia in Cirulation Researh, Volume XII, February 190S dogs appeared to be slightly different from the hvpopotassemia pattern in man and in the rabbit 14 ' 15 beause the dogs had no apparent U waves. This was noted by "Weller et al. 7 and by Young and Daugherty. 0 It appears unlikely that hanges in shape of the ventriular ation potential ourring during K depletion differ basially in these speies of mammals. Therefore, the lengthening and nothing of the T wave in hypopotassemi dogs probably reflet the same hanges in the desending limb of the ation potential -whih are responsible for appearane of U waves in the isolated rabbit hearts. 10 The ommon denominator of the CG hanges in hypopotassemia in the dog, rabbit, and man is the appearane of diastoli potentials whih hange the shape of the terminal part of the T wave and are superimposed on the U wave if this wave is present normally. An absene of U wave in normal dogs and presene of U wave in normal man annot be explained at the present time. The plateau part of the ventriular ation potential is also hanged during hypopotassemia, and this aounts for hanges in position of the S-T segment and onfiguration of. the early portion of the T wave. 10 In the present study we observed suh hanges, partiularly depression of the S-T segment but we did not inlude them in the quantitative study. In other studies the S-T depression appeared late 7 and oasionally persisted after potassium repletion. 7 " 9 The latter observation suggests that these hanges may not be as speifi as the other rapidly reversible eletroardiographi hanges. The P wave, whih orresponds to depolarization of the atrial myoardium, inreases in amplitude in hypopotassemi dogs. 7 " 9 Tu et al. 8 found a orrelation oeffiient of 0.61 between the inrease of P amplitude and the derease of plasma K and a orrelation oeffiient of 0.78 between the inrease of P amplitude and the ratio of myoardial potassium to plasma K. In our study the widening of the QRS appeared to orrelate better with the derease in K p than other CG hanges. The widening
150 SURAWICZ, KUNTN, SIMS of the QRS omplex was of a generalized type whih suggests that this is not aused by delayed ondution in one of the major bundles but by a delay in either the peripheral twigs of the onduting system, Purkinje fibers or myoardium. Although the ause of the generalized widening of the QRS omplex in hypopotassemi animals remains unexplained, one must assume that it is related to low K beause it ours onsistently not only in dogs' but in rabbits 18 and in some humans. 17 Weller et al. T found QRS widening only after serum K onentration dereased to about 2.0 mq/l while Tu et al. s found no hanges in QRS duration in their hypopotassemi dogs. The disrepany between these observations and our findings an probably be explained by the fat that the QRS widening, in spite of its onsisteny, was of relatively small magnitude and the end point of the QRS omplex is somewhat diffiult to define. We have inreased resolution of the QRS omplex by reording at a higher paper speed, by magnifiation of the traings and by employing a standard method of measurement. 3 We found no orrelation between hanges in the P-R interval and derease in K,, onentration. Similarly we found no orrelation between hanges in heart rate and derease in K p onentration in ontrast to Weller et al, T who observed an inreased heart rate during the rapid deline of the plasma K onentration. This disrepany may be due to a different type of anesthesia or to other nonspeifi fators. In spite of detailed differenes between our results and those of Weller et al. 7 we would subsribe to their general onlusion, namely, "... aute depletion of potassium faithfully ontributes to the eletroardiographi pattern of hypokalemia." Summary Hypopotassemia was produed in dogs by hemodialysis or by intravenous administration of gluose with insulin. Both methods resulted in a similar derease in plasma K onentration and similar CG hanges although K was withdrawn from the ells during dialysis while it entered the Gells during gluose-insulin infusion. The maximal plasma K dei'ease ourred after two hours of dialysis or gluose-insulin infusion; administration of gluose-insulin after dialysis resulted in no further plasma potassium derease. However, dialysis after gluose-insulin infusion aused further lowering of plasma potassium. The CG hanges were evaluated in relation to hanges in plasma K onentration. The most onsistent effets of hypopotassemia were in dereasing order: inreased duration of the QRS omplex, inreased duration of the QT C interval and inrease in amplitude of the P wave. Correlation oeffiients (R) between these CG hanges and the plasma K dereases were of the order 0.48 to 0.54. The CG hanges failed to show even a weak orrelation with absolute plasma K onentrations or with alulated hanges in intraellular K. Aknowledgment The authors aknowledge with gratitude the help of Dr. G. S. Bay, Dr.. J. Wallae, Mrs. B. Louis, and Miss M. O'Connell. Referenes 1. BYWATRS,. G. L.: Ishemi musle nerosis. J.A.M.A. 124: 1103, 1944. 2. KOLFW, W. Y.: New Ways of Treating Uremia. J & A. Churhill Ltd., London, 1947. 3. MRRILL, J. P., LVIN, H. D., SOMMRVILL, W., AND SMITH, S. Ill: Clinial reognition and treatment of aute potassium intoxiation. Ami. Int. Med. 33: 797, 1950. 4. ISRI, O-, AND MUNSON, P.: Rapid determination of alium by diret tit.ra.tion with thylendiainine tetraaetate (DTA) : Personal ommuniation. 5. LPSCHKIN,., AND SUKAWICZ, B.; The measurement of the duration of the QS interval. Am. Heart J. 44: 80, 1952. 0. XINCK,. M., HOLLAND, C.., AND STUTZ- JI.W, P. L.: Homeostasis of potassium in the extraellular fluid of the dog during removal by vivodialysis. Am. J. Physiol. 156: 290, 1949. 7. WLLR, J. M., Lows, B., HOIGN,. V., WYATT, N. F., CKISCITLLO, M., MRRILL, J. P., AND LVIN, S. A.: ffets of aute removal of potassium from dogs. Changes in eletroardiogram. Cirulation 11: 44, 1955. 8. Tu, W. H., BAIR, G. O., AND SCHLORB, P..: Removal of exess body potassium by hemodialysis. Am. J. Physiol. 199: 671, 1960. 9. YOUNG, J. V., AND DAUGHRTY, G. W.: Use of Cirulation Researh, Volume XII, February ZSSS
POTASSIUM AND LCTROCARDIOGRAM 151 the artifiial kidney in the prodution and stud}' of experimental hypokalemia. Proe. Staff Mayo Clin. 31: 357, 1956. 10. HABIB, Y. A., NICHOPOULOS, G. C, AND OVRMAN, B. B.: ffet of aute removal of potassium from the body on tissue eletrolytes. Am. J. Physiol. 193: 634, 1958. 11. RBAR, B., OMACHI, A., AND BBAR, J., JR.: ffets of gluose infusion on dog rayoardial metabolism. Cirulation Researh 7: 977, 1959. 12. LPSCHKIN,.: The role of eletrolytes in metaboli influenes on the eletroardiogram. Adv. Gardiol. 2: 189, 1959. 13. VAN BUCHM, F. S. P.: The eletroardiogram and potassium metabolism. Am. J. Med. 23: 376, 1957. 14. STTKAWICZ, B., BRAUN, H. A., CRUM, W. B., KMP,.. L., WAGNR, S., AND BLLT, S.: Quantitative analysis of the eleetroardiographi pattern of hypopotassemia. Cirulation 16: 750, 1957. 15. SURAWICZ, B., AND LPSCHKIN,.: The eleetroeardiographi pattern of hypopotassemia with and without hypoalemia. Cirulation 8: 801, 1953. 16. SURAWIOZ, B., LPSCHKIN,., HRRLICH, H. C, AND HOFFMAN, B. F.: ffet of potassium and alium defiieny on monophasie ation potential, eletroardiogram and ontratility of isolated rabbit hearts. Am. J. Physiol. 196: 1302, 1959. 17. CHRRY, J. D., AND SURAWICZ, B.: Unusual effets of potassium defiieny on the heart of a hild with ystinosis. Pediatris 30: 414, 1962. Cirulation Researh, Volume XI!, February 19SS