Citrate Toxicity During Massive Transfusion

Walter H. Dzik and Scott A. Kirkley

HE USE OF sodium citrate as the blood anti­ ing complexity required rapid transfusion of larger T coagulant for transfusion science dates from volumes of blood. In 1955 several cases of 'citric 1914 to 1915 and the almost simultaneous acid intoxication' following transfusion were publication of the work of four independent reported by Bunker.9 In the years that followed, investigators. Articles from Hustin,l Agote,2 numerous investigations were published and con­ Weil,3 and Lewisohn,4 published between May siderable controversy developed regarding the 1914 and January 1915, each reported the success­ management of citrate toxicity during massive ful use of citrated blood for human transfusion. transfusion. Recognition and therapy ofcitrate tox­ Because citrate was known to be toxic to animals, icity was enhanced by the widespread application Lewisohn carefully titrated the minimum concen­ in the 1970s of ion selective electrodes capable of tration of citrate required to prevent clotting. accurate measurement of the level of ionized Despite the demonstration by Weil that citrated Ca + +. With the development of advanced blood could be stored for several days and still be trauma care, liver transplantation, and prolonged effective,3 and the discovery by Rous and TurnerSTumer5 extensive surgical procedures in pediatrics, there that citrated blood supplemented by dextrose was has been a renewed interest in the role of citrate capable of more prolonged storage, citrated blood toxicity in the setting of ultramassive transfusion. was not quickly accepted by the general medical This review focuses on citrate toxicity during community. Though used with success in limited massive blood replacement in adults. The chemis­ trialstriais on the battlefield in World War 1,6 try of the citrate calcium interaction; the dose, transfusions with citrated blood were often distribution, metabolism,metabolism, and excretion of citrate; associated with chills and fever which were the toxic effects of citrate; and the treatment of incorrectly attributed to the citrate. Febrile citrate induced hypocalcemiahypoca1cemia are discussed. reactions were so common with citrated blood that Occasionally we have offered a personal view during the 1920s most transfusions consisted of the based on our own experience with massive rapid transfers of nonanticoagulated , transfusion during hepatic transplantation. Citrate and the use of stored citrated blood did not become toxicity is also discussed in most general reviews 10 commonplace until the mid-1930s. 7 of massive transfusion. ,1l While improvements in transfusion technology and the establishment of blood banks made the CHEMISTRY OF CITRATE administration of blood a standard procedure in the operating room, blood usage was generally limited Citric acid (molecular weight 192 daltons) is a to a few units for any given patient. Advances in ubiquitous organic compound with three ionizable knowledge of the biochemistry of citrate and cal­ carboxyl groups. With three pKs (3.14,4.77, and cium led to an improved understanding of their 6.39) all <7.4, the majority of citrate present in interaction as well as the relationship of serum the body has allaU three carboxyl groups ionized (Fig ionized Ca+ + to total serum calcium.ca1cium. The devel­ 1).l). These ionized carboxyl groups are responsible opment of citrate toxicity due to acute hypocalce­hypoca1ce­ for the major pharmacologic action of citrate, the mia was demonstrated in dogs as early as 1944. 8 binding of divalent cations. This binding is After WWoddorld War II surgical techniques of increas- accomplished by having two of the valences occupied by the divalent calcium ion. Because of the third ionized carboxy group, citrate is still From the and Tissue Typing Laboratory, New highly soluble in aqueous media even when bound England Deaconess Hospital, Department ofoj Medicine, Har­ to a divalent cation. Citrate may bind to any of the vard Medical School, Boston. metallicmetaUic divalent cations and subsequently lower Address requests to Walter H. Dzik, MD, Blood Bank and the concentration of the ionized form of that cat­ Tissue Typing Laboratory, 185 Pilgrim Road, Boston, MA 02215. ion. While most reports dealing with the effects of © 1988 by Grune & Stratton, 1nc. citrate deal with its effects on ionized calcium, 0887 7963/88/0202-0004$03.00/0796318810202-0004$03.0010 wellweU documented depressions of magnesium have

76 Reviews, Vol 2, No 2 (June), 1988: pp 76-94 CiTRATECITRATE TOXICITY DURING MURINE TRANSFUSIONTRANSFUSlON 77

nondiffusable. The nondiffusable form was found CH2COO­ to be bound to serum proteins, especially albumin. I By the 1930s most researchers in the field agreed HO-C-COO- that the diffusable portion of calcium existed in two states, bound to diffusable smallsmallligandsligands such I as lactate or citrate, and in the free or ionized CH2COO- state. 18 These three states exist in equilibrium in the plasma and it is generally agreed that ionized Ca + + is the physiologically active form. In the C6 H5 07 healthy human, approximately 47% of calcium is Fig 1. Chemical structure ofot citrate. in the ionized form, 40% bound to serum proteins (mostly(most1y albumin), and approximately 13% bound also been reported. 12,13 Citrate binds slightly to smaller ligands. 19 These proportions differ with stronger to Mg + + (formation constant 2.9 X changes in pH, protein concentration, ligand con­ 103) than it does to Ca + + (formation constant centration, and total calcium concentration. 1.88 x 103).12,14,15 Citrate is found in all human While measurement of total serum calcium cells and is an intermediary in the Kreb's citric remains useful for gross or chronic disturbances of acid cycle. Because the citric acid cycle takes serum calcium, acute changes in ionized Ca + + place within the mitochondria of a cell, tissues are often missed with these measurements, Most with a high number of mitochondria per cell (such laboratories use dye binding methods such as or­ as liver, skeletal muscle, and kidney) contain thocresolphthalein or arsenazo III dye where the larger amounts of those enzymes responsible for change in absorption by the dye is proportional to 2o20 the production and metabolism of citrate. WhileWbile the total calcium concentration. AtomicAtomie neither a common nor routine laboratory test, absorption spectrophotometryspeetrophotometry is also used and is plasma citrate levels can be measured by several an accurateaeeurate and reproduciblereprodueible method, but is rarely methods. One common method involves automated and requires earefulcareful maintenancemaintenanee and incubation of plasma with bacterial citrate lyase in standardization, While these methods have good the presence of zinc ions, The reaction produces accuracyaceuraey and precision, the difficultydiffieulty is that total oxaloacetate whichwhieh is then acted upon by malate calcium measurements may not accuratelyaeeurately reflectrefleet dehydrogenase resulting in the production of the eoneentrationconcentration of ionizedealcium calcium which is the NADH from NAD. Production of NADH is physiologically activeaetive form. measured spectrophotometrically after the serum Early attempts to determine the ionized Ca + + proteins are precipitated. The normal adult plasma relied on nomograms for estimation of ionized concentration of citrate is from 0.9 mg/dL to 2,5 Ca+ + from measured total ea1cium.calcium. In 1935, mg/dL when measured with the citrate lyase MeLeanMcLean and Hastings, in their extensive mono­ method (Table 1).l). Slightly higher levels are found graph on ealciumcalcium in body fluids, developed a in children and in patients with hepatic or renal nomogram for estimating the eoneentrationconcentration of disease. 16,1716,17 ionized Ca + +. The nomogram was based on the total ealciumcalcium and total serum protein eoneentrationconcentration MEASUREMENT OF CALCIUM at a set pH and temperature. 18 With the advent of Early in this century it was realized that calcium the more rapid ion selectiveseleetive electrodeeleetrode measure­ exists in biologicbiologie fluidsfluids in at least two forms, one ment of ionized Ca + +, this nomogram and diffusable across a dialysis membrane, the other derivations of it, have been shown to give poor 21 estimates of the ionized Ca + + levels. ,22,23 The

Table 1. NormalNormai Adult Concentrations ofot Citrate and inaccuraciesinaecuracies ofearly nomograms likely result from Ionizedlonized Calcium the assumption that pH and small ligand eoneentrationsconcentrations had little effect on ionized Ca + + Citrate Ionizedlonized Calcium eoncentration.concentration. Another outdated method for mg/dLmg/dl 0.9-2.5 4.5-5.4 determining ionized Ca + + is the method of mEq/LmEq/l 0.14-0.39 2.3-2.7 Soulier whichwhieh estimated the ionized Ca + + by its tnmol/Ltnmol/l 0.047-0.130 1.1-1.4 effe'cteffe'et on the thromboplastin time of decalcifieddeealcified 78 DZIKOZIK AND KIRKLEY plasma.p1asma. This method is poor for ionized Ca + + normal ranges depend on the individuallaboratory.individual laboratory. greater than 1.5 mmo1/Lmmol/L (3 mEq/L) and can not Estimations ofthe normal range of ionized Ca + + measure values <0.5 mmo1/Lmmol/L (l mEq/L).19 are given in Table 1. The modern era of measurement of ionized Ca + + began with the introduction of ion CITRATE DOSAGE, METABOLISM, se1ectiveselective electrodes: Initially, ion se1ectiveselective AND CLEARANCE electrodese1ectrodes were limited by inference from other Dosage and Distribution b100dblood and serum components. Present day ion se1ectiveselective e1ectrodeselectrodes operate by comparing the The 1evellevel of citrate in the bloodstream duringduńng binding of ionized Ca + + with one side of a non­ massive transfusion results from the net balance of permeablepermeab1e membrane which is specific for binding citrate dose versus citrate removal. Due to the ca1ciumcalcium ions and comparing it to a reference complexcomp1ex interrelation of severa1several factors affecting 20 solution bathing the other side of the membrane. 20 citrate delivery and removal, predicting the level If fewer Ca + + ions adhere to one side of the ofcitrate and consequently its effect on the 1evellevel of membrane than the other, an electric potentialpotentia1 is ionized Ca+ + is extremelyextreme1y difficult.difficu1t. 24 The dose set up which can be measured via a second refer­refer­ of citrate is determined by several factors including ence electrode, in contact with the serum samp1e,sample, the particularparticu1ar blood component, the anticoagu­ acting as a salt bridge. The membranes bind the lant preservative formulation,formu1ation, the rate of calcium by either an ion exchange mechanism, administration, the recipient bloodb100d volume, and where the binding moiety of the membrane forms the duration of administration. The citrate burden a ca1ciumcalcium salt, or by forming a neutralneutra1 but steri­ of various anticoagulant preservative formulations cally and electrostatically favorable binding pocket is shown in Table 2. Citrate (MW = 189) is for the calcium ion. 2o20 Both these methods give present as both trisodium citrate dihydrate (MW = measurements which are rapid, highly reproduc­ 291) and citriccitńc acid monohydrate (MW = 210) in ible, and can be adjusted to sample whole blood, most formulations.formu1ations. Given the molecularmolecu1ar weights, plasma, or serum. Measurement devices which citrate represents 65% and 90%oftrisodiumtńsodiumcitrate employ ultrafiltration or dialysis are not as useful and citric acid respectively. The concentration of because ligand bound calcium, including citrate citrate can be estimated for whole blood, red b100dblood bound calcium is also measured. 20 One drawback cells, and (FFP)(PFP) or to the ion selective electrode is the lack of a (Tab1e(Table 2). The highest concentration of citrate is standard reference method, making results be­ naturally found in FFP. Although for many years tween different devices and different laboratorieslaboratońes acid-citrate-dextrose (ACD) solution presented the difficult to compare. Work is presently being done greatest citrate doseto the b100dblood recipient, the to come to an agreement for such a standard. Thus, newest additive formulations (AS-3) have the

Table 2. Citrate Content ofot Various Anticoagulant-Perservative Formulations

ACD CPD/CPDA-1 AS-1 AS-3

Grams trisodium citrateeitrate (2H 2O) 1.485 1.656 1.656 2.244

Grams citriceitrie acid (H 2 O) .540 .206 .206 .248 Grams citrate per unit 1.451 1.261 1.261 1.681 ConcentrationConeentration ofot citrate per liter (mmoI/L) 7.6 6.7 6.7 8.9 ConcentrationConeentration of citrate (mg/dL)* in: Whole Blood 280 246 206 274 Packed RBC 87 76 54 181 FFP 436 384 384 384 Quantity of citrate (mg)* in: Whole Blood 1451 1261 1261 1681 PackedPaeked RBC 200 176 176 596 FFP 976 843 843 843

* Calculated assuming a 450 mL donation; HCT, 41, and no movement of citrate into cells.eells. For ACO,ACD, CPDCPO and CPOA-lCPDA-l assumes the production of packedpaeked RBC with Hct,Het, 80; FFP, 230 mL; and concentrate,eoneentrate, 55 mL. For AS-l and AS-3 assumes productionproduetion of red blood cellseells with final Hct,Het, 56; FFP, 230 mL; and platelet concentrate,coneentrate, 55 mL. , CITRATE TOXICITY DURING MURINE TRANSFUSlONTRANSFUSION 79 highestbighest quantity of citrate. AS-3 red blood cells rapid transfusion would be expected to take several deliver three times the quantity of citrate as ACD hours. Evidence for such delayed clearance has or CPD packed cells. been found in studies of metabolism and renal For any given ,produet, the rate of excretion of citrate following massive trans­ 28 29 administration and the size of the recipient are the fusion. 28.29• key determinants of citrate administration. Thus, The duration of citrate administration is also a studies of citrate toxicity generally refer to mg ci­ key determinant ofthe blood citrate level. Platelet­ tratelkg recipient/minute. For example, one unit of pheresis studies with relatively constant low dose CPDA-lCPDA-1 whole blood administered to a 70 kg man citrate infusion over approximately 100 minutes over five minutes corresponds to 1261 mg citrate/ have shown that the citrate level attained is only 70 kg/5 min or 3.6 mg citratelkg/min. Citrate is 25% of that which would be predicted from rapidly removed by the liver and kidney. How­ redistribution throughout the extracellular space. ever, the distribution of citrate also plays an Without metabolismmetab01ism and excretion ofcitrate during important role in determining the citrate level fol­ the period of infusion, toxic levels ofcitrate would 30 lowing transfusion. Although the relative have developed. ,31 In two independent studies importance of metabolismmetabolism v redistribution of ci­ very similar blood citrate levels developed in trate is not fullyfu1ly investigated, citrate can be con­ individuals receiving 4 mg citrate/kg/min over five sidered as a first order approximation to be distrib­ minutes as were found in individuals receiving uted throughout the extracellular fluid space. This only 1.6 mg citrate/kg/min over 113 minutes. 30 distribution occurs within five minutes of infusions Thus, prolonged rapid would be of mild-to-moderate quantities of citrate. For ex­ expected to result in higher citrate levels than ample, in one study 500 mL of citrated blood was equally rapid infusions of short duration. This is of infused over five minutes to adults (4 mg citrate/ clinical importance in settings such as hepatic kg/min x 5). Citrate levels were measured every transplantation where rapid transfusion may minute. At the end of the infusion, the measured continue for hours. With prolonged duration of citrate level was 66% that which would have been blood administration, citrate metabolismmetab01ism and predicted had citrate remained in the intravascular excretion (rather than distribution) become the space. Within three minutes after stopping the mostimportant defense against the development of infusion, the blood level was equal to that which toxic concentrations of citrate. would have been predicted had citrate distributed Another means for the body to deal with a low itself over the extracellular volume. 25 After ionized calcium due to a citrate load is to mobilize infusion of 2.35 mg citrate/kg/min over ten stores of calcium. Parathyroid hormone (PTH) minutes, the concentration of citrate at the end of levels are measurably elevated within 2 to 4 the infusion was equal to that which would have minutes after administration of citrate32 and reach been expected from redistribution over the extra­ peak levels between 5 and 15 minutes after cellular volume. 9 However, rapid challengeschallenges infusion. 13 Elevations in PTH have been found of large quantities of citrate can exceed both experimentally and during surgical pro­ redistribution, metabolism, and excretion. 26 Five cedures requiring blood support in adults and patients receiving a mean of 5.5 mg citrate/kg/min during exchange transfusions in infants. 13 During over 15.6 minutes developed average peak citrate , one study found that the ionized levels of 62 mg/dL which was 1.5 times greater calcium remained low with continuous citrate than the expected citrate level assuming complete infusion but the total calcium dropped at first, then extracellular redistribution. 27 As a result of metab­ returnedretumed to near normalnormallevels.levels. UltrafilterableUltrafilterabie cal­ olism and redistribution outside the vascular space, cium, which equals the total ionized Ca + + plus there is an initial rapid exponential decline in the ligand bound Ca + + (but not the protein bound concentration of citrate after cessation of rapid calcium) continued to rise during the blood infusion. Further removal of citrate results procedure. 13l3 This is probably a result of calcium from continued metabolism and renal excretion. mobilization with most ofthe newly mobilized cal­ Due to the permeability of cells to citrate and the cium being bound to the infused citrate. Protein apparent large volume of distribution, complete bound citrate decreased by 35% during the metabolismmetabolism and excretion following prolonged procedures, representing a significant buffer of 80 DZIK AND KIRKLEY calcium when ionized Ca + + levels drop. These mitochondria. Once insideinside the mitochondria, two changes may be affected by temperature, pH and carbons of citrate are removed and two molecules other alterations in addition to the variability of ofC0of CO2 formed. With additional turnstums ofofthethe Kreb's various PTH assays among different methods and cycle, the original carbons of citrate are converted 33 laboratories. to CO2 •, Since turning of the Kreb's cycle regenerates cycle intermediates from the carbons Citrate Metabolism introduced by acetylCoA, metabolism of exoge­ The metabolism ofcitrate involves multiple bio­ nous citrate exclusively by Kreb's cycling would chemical pathways (Fig 2). Citrate can directly not reduce the concentration of citrate nor have a enter the Kreb's tricarboxylic acid cycle to be com­ net effect on acid/base balance. Instead, the Kreb's pletely metabolized to CO2 and H20, can partici­ cycle intermediates are able to be metabolized by pate in fatty acid and amino acid synthesis, and can additional pathways. Within the mitochondria, ox­ be converted to glucose via gluconeogenesis. Each aloacetate can be converted to pyruvate by the of these pathways will be reviewed in tum.turn. action of pyruvate carboxylase, which under Overall, the complete oxidation ofcitrate results in normal conditions serves to convert pyruvate to the production of CO2 and H20. Complete metab­ oxaloacetate in an anapleurotic reaction. olism of ionic citrate - 3 to nonionic endproducts Conversion to pyruvate requires biotin as a cofac­ consumes three H + ions according to: tor, consumes a second H + ion, and yields CO2 ,• 3 The pyruvate formed can be then converted to ace­ C H 0 - + 4.5 O + 3 H+ ~ 4 H 0 6 j 7 2 2 tylCoA in a seriesseries of reactions that consume the + 6 CO2 third H + ion and yield an additional CO2 ,• Exogenous citrate can be actively transported However, other Kreb's cycle intermediates are through the mitochondrial membrane to participate also able to be transported through the mitochon­ in Kreb's cycle reactions. Tricarboxylic acids (ci­ drial membrane. Alpha-ketoglutarate and malate trate, isocitrate, aconitate) are actively transported are two such intermediates. As the concentration across the mitochondrial membrane by carrier of cycle intermediates increases inside mitochon­ molecules. Movement ofcitrate into the mitochon­ dria following citrate administration, alpha-keto­ dria is linked to movement of malate out of the glutarate and malate would be expected to leave

CITRATE METABOLlSMMETABOLISM

Exogenous Ctlrofe-3 Exogenous Ct~rote-3

g 2 SY~f~S'SSy~f~S'S \ Asport]ote.,ASPort]ote.,a.Ky lutorot'-lutorote

ł C Glutomot.-'Glutamate' Acetylte oA 2 0,0100celol..- ~------

Fig 2. Metabolism ofot citrate. CITRATE TOXICITY DURING MURINE TRANSFUSIONTRANSFUSlON 81

the mitochondria. Evidence suggests that metabo­ increased cytoplasmic citrate as a result of massive lism ofexogenous citrate results in net transport of transfusion might be expected to temporarilytemporańly stim­ malate outside the mitochondria.mitochondńa. Because the me­ ulate fatty acid synthesis. The cytoplasmic oxalo­ tabolism of citrate to malate involves incomplete acetate formed as a result of citrate cleavage may turningtuming of the Kreb's cycle, the normal balance of then be further metabolized according to the H + ion production and consumption is not reactions outlined above. The metabolism of cyto­ maintained and a net single H + ion is consumed. plasmic citrate via citrate cleaving enzyme to ox­ anceOnce in the cytoplasm, malate is able to be aloacetate and then to PEP, pyruvate, and mito­ converted to oxaloacetate which can participate in chondrial acetylCoA also consumes three H + two biochemical pathways described below. ions. Oxaloacetate is a ketoacid and as such can The rate limiting step of citrate metabolismmetabolism fol­ undergo a transaminase reaction with glutamate to lowing massive transfusion is unknown. Evidence form aspartate and alpha-ketoglutarate. The reac­ from liver transplantation and from studies ofrenal tion is reversible with no net loss of oxaloacetate excretion of citrate suggest that alkalemia slows carbons and no net production or consumption of the metabolismmetabolism of citrate presumably by retarding H+ ion. A second pathway of cytoplasmic oxalo­ movement of citrate and malate across the mito­ acetate metabolism involves conversion to phos­ chondrial membrane. Whether factors which might phoenolpyruvate (PEP) by PEP carboxykinase. influence the activity of citrate cleaving enzyme This enzyme is a key enzyme in gluconeogenesis exert an overall effect on the metabolismmetabolism of and is found in cellscelIs such as those of the liver, administered citrate is less wellwelI studied. kidney, and skeletal muscle. The reaction requires energy in the form of GTP and results in the Citrate Clearance release of CO2., Thus, exogenously administered Clearance of citrate is highest in those organs citrate can be converted to PEP through oxaloace­ which receive a high proportion of the cardiac tate. Onceance formed, PEP can be converted to output and which are composed of cellscelIs with nu­ glucose via the enzymes ofgluconeogenesis. How­ merous mitochondria.' CellsCelIs which are dependent ever, during conditions of AATPTP depletion the me­ on glycolysis for their energy needs, such as red tabolism of PEP to pyruvate is favored. The blood cells,celIs, have low levels of citrate and do not formation of pyruvate from PEP consumes an remove citrate from the circulation. Liver, kidney, additional H + ion. Pyruvate is freely permeable to and skeletal muscle are responsible for most of the the mitochondria.mitochondńa. anceOnce inside the mitochondria, it metabolism and excretion of citrate. can be converted to acetylCoA in a series of Since the 1930s, investigators have recognized reactions that release CO2 and consume one H + that basal serum citrate levels are slightly higher in ion. The acetylCoA is then free to combine with patients with chronic liver disease than in those mitochondrial oxaloacetate and be metabolized to without liver disease. 16 ,34 In normalnormaI fasting CO2 and H20. Thus, the complete metabolic humans, about 20% of the endogenous citrate in breakdown of the six carbons of citrate maymayaIsoalso the serum is removed with each pass through the occur via pathways that include extramitochondrialextramitochondńal liver. 16,3416,34 In one study when a perfusate containing enzymes, produce CO2, and consume three H + 54 mg/dL citrate was presented to isolated calf ions. Citrate conversion to glucose via gluconeo­ livers, there was a rapid initial clearance of 50% of genesis yields fewer CO2 molecules but also the citrate over the first thirty minutes followedfolIowed by consumes three H + ions. a more gradual clearance of the remaining 50% An additional biochemical pathway of citrate over the next two and a half hours. 35 Convincing metabolism is that provided by citrate cleaving evidence of the importance of the liver in citrate enzyme (Fig 2). This enzyme, ATP-citrate lyase is clearance in humans is seen during liver found in liver and adipose cellscelIs and splits cytoplas­ transplantation when there is a marked rise in mic citrate in the presence of CoASH to form ace­ serum citrate with concomitant drop in ionized tylCoA and oxaloacetate. One H + ion is Ca + + during the anhepatic phase of the surgery consumed in the reaction. The acetylCoA formed with a rapid reversal of these changes when the is not permeable to the mitochondria,mitochondńa, but is able to new liver is perfused.22.36.37 Citrate clearance by participate in the biosynthesis of fatty acids. Thus, the liver appears to be greatly reduced during hy- 82 DZIK AND KIRKLEY pothennia, increasing as the patient is warmed. the urine,42 Up to 60% of filtered citrate may Hypotension with decreased hepatic perfusion also appear in the urine of humans during metabolic leads to a decrease in clearance of citrate by the alkalosis,41 This increase in excretion results from liver. 28 inhibition of citrate transport into mitochondria by The kidney is also essential for citrate clearance. the tricarboxylictriearboxylic acid carrier. Other conditions Some investigators feel that the kidney is the most leading to increased urinary excretion of citrate important organ for citrate clearance since it can include high levels of other organic acids of the metabolize large amounts of citrate and also can Krebs' cycle such as fumarate or malate and spe­ excrete nonmetabolized citrate in the urine. 38,39,4O cificcifie inhibitors of the Krebs' cycle pathways such Estimates of the amount of exogenous citrate as fluorocitrate and malonate, Calcitonin, vitamin handled by the kidney range from 20%28,36 to D, calcium, and magnesium have also been shown 41 68%40 of a given load in humans. Most of the to increase citrate excretion. 41 Conversely, the uptake ofcitrate by the kidney is from reabsorption excretion of citrate is reduced during acidosis. 42 of filtered citrate in the proximal tubule but up to The effects of K + and bicarbonate levels on ci­ 30% of the total renal uptake is peritubular uptake trate excretion appear to be due to their effects on from postglomerular blood.41 The proximal acid base balance rather than a direct effect of the 42 tubular cells, which are rich in mitochondria, are ion. In the clinicalclinieal setting, clearance of citrate by responsible for the metabolism of citrate. the kidney may also be limited by the decreased Normally < 1%I% of the citrate filtered by the kidney glomerular filtration rate and renal ischemia that is excreted in the urine, but in the presence of high frequently develops during massive transfusion. plasma concentration of citrate or alkalemia, there More work is needed to assess the relative is a dramatic increase in the fractional excretion in contributions of the liver and kidney in citrate me-

+

;-.C'

+J:r':

Fig 3. Effect of rapid citrate infusion on the EKG. The top panel shows a prolonged GoTcQoTc interval. Following a period of rapid transfusion, the ionizedlonized calcium fellfeli to 0.6 mEq/L and a widened GRSQRS complex devel­ oped Imiddle(middle panell.paneli. The effect was reversed with intravenous administration of calcium (bottom panel). CITRATE TOXICITY DURING MURINE TRANSFUSlON 83 tabolism and excretion during massive citrate increases more sharply at severely depressed levels loads. of ionized Ca + + (0.25 mEq/L to 1.0 mEq/ L).45,51 A study of twenty adults undergoing in­ TOXICITY OF CITRATE traoperative rapid transfusion correlated corrected QT intervals and ionized Ca and found the Effects on the EKG + + correlation coefficient was only r = .61. 45 The Depression of the ionized Ca + + has correlation was slightly higher using QoTc characteristic effects on the EKG. The most compared with QTc. The 95 percent confidence commonly recognized effect is prolongation of the intervals indicated that a given QoTc was often QT interval (Fig 3). The QT interval corresponds associated with a twofold or greater range in to the time from contraction to repolarization and predicted ionized Ca+ + levels.45 Because the varies with the heart rate. The corrected QT range of ionized Ca + + associated with any given interval (QTc), which takes into account the effect QTc interval is broad, measurement of the QTc of heart rate, is the time from the origin ofthe QRS interval cannot be substituted for accurate mea­ complex to the end of the T wave divided by the surement of the ionized Ca + + in patients likely square root of the RR interval. The corrected QT to develop citrate toxicity during massive interval mayaIso be defined by substituting the transfusion. time from the origin of the QRS complex to the The effect of citrate infusion on the QT interval origin of the T wave rather than the end of the T has also been studied in the setting of apheresis wave, and is then referred to as the QOTC. 43 The procedures uncomplicated by many of the QTc in normal adults ranges from 350 msec to 440 confounding variabIes found during intraoperative msec44 and the QoTc from 180 msec to 240 massive transfusion. One study of 15 apheresis msec. 45 The measured QT interval for a particular procedures measured a 32% average decrease in individual is influenced by numerous factors the level of ionized Ca + + with a corresponding including age, sex, autonomic innervation, myo­ prolongation of the QT interval by an average of cardial ischemia, antiarrhythmics, hypothyroid­ 80 msec (range 50 msec to 120 msec).30 These ism, and severe hypothermia in addition to effects occurred during an average citrate infusion .46 of 1.58 mg/kg/min for 113 minutes resulting in a The relationship between hypocalcemia and a mean postprocedure citrate level of 26.7 mg/dL. prolonged QT interval has been recognized for Although the QT interval correlated with the rate many years. In early studies, sodium citrate was and dose of citrate administered, the level of infused into normaI conscious volunteers or ionized Ca + + could not be predicted from the patients under anesthesia and the prolongation of QT interval. Following cessation of the procedure, the QT interval observed. 26,27,47 Similar studies in the QT interval returned to normal in 2 to 15 dogs documented similar effects on the EKG and minutes which was significant1y less than the time demonstrated that the effects were abolished with required for levels of ionized Ca + + and citrate to administration of calcium.48 ,49,50 Despite the re­ return to normal. In another study of 12 platelet­ producible effect of citrate on the QT interval, pheresis procedures, prolongations in the QT, studies in patients undergoing transfusion showed QTc, QoT, and QoTc were correlated with the that the QT interval was an unreliable guide to level of ionized Ca+ + during procedures that administration of supplemental calcium. 24 The re­ infused citrate at an average rate of 1.38 mg/ lationship between the level of ionized Ca + + and kg/min. 52 Although the best correlation was found the length of the QT interval during rapid with the QoTc (r = .59), the QoTc could not transfusion was found to be nonlinear and has been predict the level of ionized Ca + +. Among described by both 10garithmic51 and hyperbolic45 individuals with depressed levels of ionized curves. At mild depressions of the ionized Ca + + Ca + +, only 9 of 30 QoTc measurements were (>1.75 mEq/L), little or no effect on the QoTc is longer than the longest baseline reading. In seen. At more moderate depressions of ionized addition, the authors found that the level ofionized Ca + + (1.0 mEq/L to 1.75 mEq/L), the average Ca + + could not be predicted by the degree of QoTc interval for a group of individuals will prolongation of the QT interval over baseline for a become prolonged.45 ,51 This prolongation given individual. In a third study of 79 platelet- 84 DZIK AND KIRKLEY pheresis procedures levels of ionized Ca + + the duration of the action potential plateau. A ranging from 2.25 to 1.35 mEq/L were found to be reduction in extracellular ionized Ca + + results in in 1inearlinear relation to levels of serum citrate ranging a decrease in the slow outward K + current and from 2 to 40 mg/dL. 31 However, the average QTc prolongation of the plateau. Because the action prolonged by only 13 msec following the potential plateau without the T wave is measured procedure. by the QoTc, the QoTc correlates better with the When ionized Ca + + levels become severelysevere1y degree of hypocalcemia than the QTc. Whether depressed, other effects on the EKG are observed. citrate anions have any independent effect on my­ EarlyEar1y reports of ventricular fibrillation and cardiac ocardial permeability to Ca + + or K + and exert arrest during massive transfusion lack complete any effect on the action potential plateau indepen­ documentation of electrolyte and acid/base dent of direct lowering ofthe extracellular concen­ disturbances. 53 ,54,55 Progressive severe hypocalce­ tration of ionized Ca + + is not well understood. miarnia would be expected to result in prolongation of the QRS complex. Fig 3 shows an intraoperative Effects on Ventricular Performance EKG from an adult patient in our hepatic In 1883, Sydney Ringer demonstrated that the transplantation program. Before the onset of rapid isolated frog heart was capable of sustained con­ transfusion, the tracing showed anormaIa normal QRS tractions when suspended in saline. 57 One year morphologymorpho10gy but a prolongedpro10nged QoTc interval. In the later, Dr Ringer published a second paper in the middle panel is shown the EKG during a period of same journaljoumal noting that the water used to prepare rapid blood infusion with a corresponding pH = the saline of his previous study was not distilled 7.26, K+ = 5.2, Temperature = 92°F and water, but rather pipe water obtained from the New ionized Ca + + = 0.6 mEq/L. Following two Water Company, London. 58 Chemical analysis grams of intravenous (IV) calcium chloride and no showed it to be contaminated with appreciable change in pH, K +, or temperature, the EKG quantities of calcium. Ringer repeated his returnedretumed to normal (bottom panel). Such dramatic experiments and discovered that sustained beating effects on the EKG are rare outside the setting of of the heart was dependent on the presence of ex­ liver transplantation or rapid blood exchange in tracellular calcium. In the century that has neonates. followed, an enormous body of information has The mechanism of prolongation of the QT developed which documents the importance of cal­ interval during citrate toxicity and hypocalcemiahypocalcernia is cium in myocardial performance. An excellent likely related to the plateau phase of the myocar­ review has been recently published.59 dial action potential. 56 The action potential of ven­ For over thirty years it has been recognized that tricular depolarization/repolarization is divided massive transfusion with citrate toxicity and hy­ into discrete phases which are regulatedregu1ated by pocalcemia would be expected to decrease car­ selective movement of ions across the myocardial diac performance and early studies in both cell membrane. The initial rapid depolarization humans47.6o.61humans47.60.61 and animals49 ,62.63.64 supported the results from a closure of K + channels and an concept. Nevertheless, acceptance of citrate toxic­ opening of fast moving Na + channels. This ity was not without some degree of controversy. 26 corresponds temporally to the QRS tracing on the The development of cardiac surgery and radical EKG. Depolarization is then sustained by a bal­ cancer surgery in the 1950sprompted several in­ anced slow inward movement ofCa + + and Na + vestigations into the cardiovascular effects of ci­ and a slow outward movement of K +. Gradually trate. In an early study,Bunker examined the the potential across the membrane becomes more effect of transfusion on blood pressure and on negative as K + ions leave the cell. anceOnce a thresh­ serum levels of citrate and total calcium.9 Twen­ old is reached, fast K + channels open and a sud­ ty-four patients receiving a median of 3,500 mL of den movement of K + outside the cell repolarizes blood developed serum citrate levels of 10 mg/dL. the membrane. Myocardial repolarization is less The median rate of citrate infusion was 0.92 mg/ well coordinated for the aggregate of myocardial kg/min (range 0.3 to 6). In 10 patients, systolic cells than depolarization and, therefore,results in a blood pressure fell to <100 mmHg. In a subse­ broad wave of repolarization (T wave). The dura­ quent more detailed study in 1962, Bunker exam­ tion of the QT interval is, therefore, dependent on ined the cardiovascular effects of direct citrate CITRATE TOXICITY DURING MURINE TRANSFUSIONTRANSFUSlON 85 infusions into six lightlylight1y anaesthetized adults atrial pressure, animals receiving an equal volume undergoing stripping of leg varicosities. 27 The of citrated blood incompletely restored aortic patients were challenged with citrate at rates blood pressure and cardiac output and developed ranging from 3.7 to 7.4 mg/kg/min for 9 to 19 considerable elevation in left atrial pressures. The minutes. Recipients developed citrate levels of 30 more rapid the rate of citrated blood infusion, the to 77 mg/dL. In 5 of 6 individuals, there was a greater the depression in left ventricular perfor­ decline in stroke volume and left ventricular work, mance. Four of nine dogs receiving citrated blood with a corresponding rise in pulse. Mean arterial at maximalmaximaI rates of 6 to 9 mLlkg/minmL/kg/min suffered pressure in these patients decreased 26%. The car­ cardiac arrest.arresL Thus, the overall effect of diovascular abnormalities corrected promptly with transfusion with citrated blood was to blunt the stopping the citrate or with infusion of calcium normalnormaI left ventricular response to volume chloride. The one patient who developed no mea­ loading. surable cardiovascular changes was the patient The blunting of the left ventricular response to who received the lowest citrate challenge (3.7 mg/ volume loading resulting from the rapid kg/min) and who developed the lowest blood ci­ administration of citrated blood was conclusively trate level (30 mg/dL). In the same report, a sim­ demonstrated in man in a well designed 1976 ilar effect was observed in dogs at citrate levels of study.67 Nine patients undergoing coronary revas­ 50 to 75 mg/dL. Serious cardiovascular depression cularization were studied. Each had normal ven­ or death occurred in the animals at citrate levels of tricular function with ejection fractions >70 $0 to 190 mg/dL and rates of citrate infusion of 10 percent.percent. Before , the to 15 mglkg/minute. Subsequent studies in dogs by patients were transfused with two units of citrate­ other investigators showed similar overall phosphate-dextrose (CPD) blood. Each unit was results. 48,50,65 37 C, pH 7.4 and <48 hours old. One unit was During the 1970s knowledge of the effects of recalcified and heparinized and the other was hep­ citrate on left ventricular performance was refined. arinized but not recalcified. Each patient served as Rather than infusing citrate, these studies involved his own control. The order of transfusions was infusions of citrated blood versus recalcified ci­ randomized and blinded to those recording cardio­ .trated blood to which heparin was added. In one vascular measurements. Each unit of blood was study of six patients undergoing open heart transfused over three minutes (2 mL/kg/min for 3 surgery,66 infusion of warm citrated blood at 150 min). After the first transfusion was administered, mLiminmL/min for three minutes resulted in a 28% the equivalent volume of blood was removed, the decrease in cardiac output with no consistent patient allowed to return to a stablestabIe state, and the change in mean arterial blood pressure, but with a second unit infused. Cardiovascular measurements rise in left atrial pressure. In contrast, the infusion were taken every 45 seconds during the of heparinized, recalcified, citrated blood at the transfusions. The results from this study are shown same rate resulted in an 18% mean increase in in Figs 4 and 5. Although patients receiving one cardiac output. Although administration of 200 to unit of citrated blood in three minutes increased 300 mgm of (IV) calcium chloride had little effect left ventricular performance in response to vol­ on cardiac output prior to the infusion of the ci­ ume, the magnitude of increase was blunted trated blood, the same dose of calcium admin­ compared to those individuals receiving recalcified istered after the transfusion caused a mean heparinized blood. This blunted response to vol­ increase of 24% in cardiac output. In the same ume loading was accompanied by a 27% decrease study, the authors bledbIed dogs to a mean systolic in the level of ionized Ca + + at the end of the arterial pressure of 50 mmHg which was three minute transfusion. maintained for 15 minutes. Infusion of citrated Several factors appear capable of rendering the blood at rates ranging from 2 to 3 mLlkg/minmL/kg/min to 6 myocardium more sensitive to the depressant to 9 mLlkg/min was compared with the infusion of effects of rapid infusions of citrated blood. Aci­ recalcified, heparinized blood at 6 to 9 mLlkg/min.mL/kg/min. demia, hyperkalemia, and hypothermia have all Although the infusion of recalcified blood been recognized to increase susceptibility to the consistently resulted in an increase in aortic cardiodepressant effects of citrated transfusions. pressure and cardiac output with little rise in left Following earlyearIy studies of cardiac autotransplanta- 86 DZIK AND KIRKLEY

formance in the setting of hepatic trans­ plantation.72,37,36,22,73 Patients undergoing hepatic transplantation are uniquely susceptible to the de­ velopment ofcitrate toxicity. Liver transplantation frequently involves rapid, massive transfusion of prolonged duration. A large proportion of the blood support is usuallyusualIy in the form of fresh frozen plasma with its high citrate concentratio~. T~e 2 3 .. metaboh~e Cl~­ b. LAP (mmHg) patients have decreased ability to Fig 4. Effect of citrate on left ventricular performance. rate during the phase of surgery when the hver IS There is a blunted response to volume challenge with citrated absent (anhepatic phase). Finally, liver transplant blood compared with recalcified blood. Reprinted with patients frequently develop other metabolic abnor­ permission.67 malities known to exacerbate citrate toxicity; hy­ pothermia from extensive exposure of viscera, hy­ tion in dogsdogs which suggested that denervation perkalemia from cellularceHular release by the grafted rendered such animals more susceptible to the liver, acidemia from decreased tissue perfusion effects of citrate,68 a study was done which docu­ and inability to metabolize lactic acid, and in some mented that the decline in left ventricular perfor­ patients a decreased ability to mobilize calcium mance with citrated transfusions was even worse in from bony stores as a result of preoperative hepatic animals that had been pretreated with a beta osteodystrophy. One study noted hypotension in blocker (propranolol).69 Albumin infusions which association with elevated central cardiac pressures can further chelate ionize Ca + + were shown to and response to calcium treatment in two of eleven have a minor deleterious effect on the level of transplant recipients. 37 A recent detailed stud~ of ionized Ca+ + and on cardiac performance in a nine adults undergoing liver transplantatIOntransplantatlOn group of patients resuscitated after trauma.70 monitored cardiac performance with Swan-Ganz Regional or global myocardial ischemia results in a catheterization. 36 Median citrate levels rose to 113 decrease in left ventricular performance during mg/dL during the anhepatic phase of surgery. Cal­ hypocalcemia. 71 A variety of drugs including an­ cium supplementation (mean 4.3 gm) was esthetic agents and calcium channel blockers are administered to prevent a fallfalI in ionized calcium likely to also worsen the myocardial depression of below approximately 50% of baseline thus hypocalcemia. maintaining ionized Ca + + >.56 mmollL. Recently there has been renewed interest in During the period of peak citrate effect, there was the effects of massive transfusion on cardiac per- a significantsignifieant decline in left ventricular performance (cardiac index, stroke index, and left ventricular stroke work index) without any significant change

175 in left or right cardiac filling pressures or systemicsystemie o HeparinHeporin vascular resistance. The period of left ventricular 150 / 150 cc/minee/min depression was also characterized by hypothermia ,P p (33.6C). However, myocardial contractility was 125 fi ,ció corrected by calcium supplementation without any a: . '"o0 100 change in body temperature. FollowingFolIowing revascu­ 3= i.~ u • larization of the new hepatichepatie graft, the levels of