Intermittent Warm Blood Cardioplegia
Background Warm heart surgery implies continuous perfusion with normothermic blood cardioplegia. Interruption of cardioplegia, however, facilitates construction of distal coronary anastomoses and is the method practiced by many surgeons. To determine whether intermittency is harmful, we present results from 720 coronary bypass patients, protected with intermittent antegrade warm blood cardioplegia, that were derived from a previous study of normothermic versus hypothermic cardioplegia.
Methods and Results Mean±SD age was 60.8±9.0 years; 27% of cases were urgent; 16% of patients had >50% left main stenosis, and 19% had grade III or IV ventricles. A mean of 3.2±0.9 grafts was constructed. The average aortic cross-clamp time was 61.8±22.2 minutes. The longest single time off cardioplegia (LTOC) averaged 11.4±4.0 minutes per patient. The cumulative time off cardioplegia as a percentage of the cross-clamp time (PTOC) was 48.2±18.6% per patient. LTOC and PTOC were divided into quartiles (LTOC, <10, 10 to 11, 12 to 13, and >13 minutes; PTOC, <36%, 36% to 49%, 50% to 62%, and >62%) and related to the prespecified composite outcome of mortality, myocardial infarction according to serial CK-MB sampling, and low-output syndrome (LOS). Longer LTOC was harmful (event rates per quartile, 13.5%, 10.3%, 10.9%, and 19.0%; P=.046), whereas longer PTOC was protective (16.1%, 17.2%, 9.4%, and 10.6%; P=.07). Stepwise logistic regression was performed, controlling for demographic and angiographic predictors. In the multivariate models, LTOC remained detrimental (P=.07) and PTOC remained beneficial (P=.053). Additional modeling after entering surgeon identity (P<.001) into the risk equation eliminated the PTOC effect, whereas LTOC remained predictive of adverse outcomes (P=.053; odds ratio, 1.06; 95% CI, 1.00, 1.13).
Conclusions The data indicate that a reasonable margin of safety exists with intermittent, antegrade warm blood cardioplegia. Repeated interruptions of warm blood cardioplegia are unlikely to lead to adverse clinical results if single interruptions are ≤13 minutes.
Warm heart surgery has been proposed as an alternative to conventional methods of myocardial protection.1 2 Prospective randomized clinical trials conducted in coronary bypass populations consistently report that perioperative outcomes are improved with warm blood compared with cold blood cardioplegia when such cardioplegia is administered by antegrade techniques.3 4 5
Warm heart surgery is defined as continuous coronary perfusion with normothermic blood cardioplegia. For coronary bypass surgery, interruption of either global or regional cardioplegia delivery enhances visualization during construction of the distal coronary anastomoses. As a result, warm blood cardioplegia is usually administered in an intermittent rather than a continuous manner for myocardial revascularization.3 4 5 Antegrade warm blood cardioplegia, as currently practiced, is therefore associated with periods of normothermic ischemia (time off cardioplegia, TOC).
We previously reported3 on a randomized clinical trial involving 1732 patients that compared normothermic and hypothermic blood cardioplegia. The warm treatment arm of that trial constitutes the study population for the present investigations. The prespecified primary outcome measures for the warm versus the cold trials were mortality and myocardial infarction, according to computerized ECG readings, and secondary end points were enzymatically defined: myocardial infarction and low-output syndrome. In the previous study, mortality (1.4% versus 2.5%, P=.12), enzymatically defined myocardial infarction (12.3% versus 17.3%, P<.001), and low-output syndrome (6.1% versus 9.3%, P=.01) decreased in the warm patients, whereas ECG-determined myocardial infarctions were similar in the two groups (P=.53). The objective of the present study is to characterize the relation between the intermittent administration of warm blood cardioplegia and the frequency of adverse perioperative events (mortality, enzymatically defined myocardial infarction, and low-output syndrome).
Patients (n=1732) undergoing isolated aortocoronary bypass surgery between November 5, 1990, and December 31, 1992, were enrolled in a randomized clinical trial of warm blood versus cold blood cardioplegia.3 Exclusion criteria for the study included cerebrovascular disease (internal carotid artery stenosis >80%) and renal insufficiency (creatine >200 mmol/L). All patients signed a consent form approved by the institutional ethics review committee. The study was conducted at the three adult cardiac surgical centers of the University of Toronto. There were 12 participating surgeons contributing 18 and 316 cases each. Randomization was instituted in the operating room by the sealed-envelope method. Randomization was stratified for urgent versus elective operation and by surgeon. Overall, 77% of eligible patients were enrolled. Details of the trial method have been provided.3
Patients (n=860) were randomized to receive warm blood cardioplegia in the trial. For the present investigation, we excluded the 79 patients who received cardioplegia by retrograde delivery. In addition, 61 patients were crossed over to hypothermic cardioplegia either because of excessive coronary flooding or difficulty achieving or maintaining cardiac arrest and were excluded from this analysis. The remaining 720 patients who received only antegrade warm blood cardioplegia form the basis for this study.
Cardiopulmonary bypass was instituted with a single, two-stage right atrial cannula, an ascending aortic perfusion cannula, and an ascending aortic cardioplegia cannula–vent line. Standard cardiopulmonary bypass management included membrane oxygenators, arterial line filters, nonpulsatile flows of 2.4 L · min−1 · m−2, mean arterial pressure >50 mm Hg, moderate hemodilution (hematocrit, 20% to 25%), and α-stat acid-base balance. Target systemic temperatures were between 33°C and 37°C.
Blood cardioplegia was prepared by mixing four parts of oxygenated blood with a crystalloid additive by use of commercially available delivery systems and was administered at 37°C. The crystalloid additive (Fremes’ solution) was prepared locally by the hospital pharmacy in high- and low-potassium formulations (high potassium: KCl 100 mmol/L, THAM 12 mmol/L, MgSO4 9 mmol/L, dextrose 250 mmol/L, CPD-adenine 20 mL) (low potassium: KCl 30 mmol/L, THAM 12 mmol/L, MgSO4 9 mmol/L, dextrose 250 mmol/L, CPD-adenine 20 mL).
High-potassium cardioplegia (1000 to 1500 mL) initially was administered at 200 to 300 mL/min into the aortic root at an aortic pressure of 80 mm Hg to achieve cardiac arrest, and low-potassium cardioplegia subsequently was infused. If cardiac activity recurred during the cross-clamping interval, high-potassium cardioplegia was reinstituted. Warm blood cardioplegia was delivered in a continuous manner at 50 to 150 mL/min, provided that cardiac distention or coronary flooding did not occur. Warm blood cardioplegia was usually interrupted (<10 to 15 minutes) to facilitate coronary visualization during construction of the distal anastomoses. In 80% of patients, cardioplegia was stopped for >5 minutes. After interruption, warm blood cardioplegia (>500 mL) was delivered at 200 to 300 mL/min, after which the rate was reduced to 50 to 150 mL/min.
Distal anastomoses were constructed with reversed saphenous veins, initially followed by free arterial conduits and then by in situ arterial conduits. Each of the surgical techniques used in the trial allowed for vein graft infusion of cardioplegia to improve distribution of the solution beyond coronary stenoses. Cardioplegia was administered through completed vein grafts by use of a manifold system. Proximal anastomoses were performed according to the surgeon’s preference after completion of the distal anastomoses before aortic declamping (21.7%) or after aortic declamping with a partial occlusion clamp (57.4%). Alternatively, distal and then proximal anastomoses were constructed sequentially (18.9%). The remainder of the patients had in situ arterial grafting without proximal anastomoses.
Data relating to cardioplegia volume administered and the length of interruptions were collected prospectively. Intermittency was operationalized according to the longest single ischemic time in minutes per patient (longest time off cardioplegia, LTOC). Intermittency also was summarized as the total duration of ischemic times or cumulative ischemic time as a proportion of the cross-clamp time per patient (percentage of time off cardioplegia, PTOC).
The primary outcome of this study was prespecified as the composite end point of mortality, myocardial infarction by enzyme criteria, and low-output syndrome.
Operative mortality included all-cause postoperative mortality to 30 days or in-hospital deaths for patients hospitalized >30 days.
Serial blood samples for creatine kinase–MB (CK-MB), obtained postoperatively at 0, 4, 8, 12, 20, and 28 hours, were analyzed in duplicate at a core laboratory; discrepancies between duplicates >10% were re-run. Mean levels were used for analysis. The area under the curve was determined for patients with five or six samples. Thresholds for enzymatic infarction (cut point, 645 IU · h) were defined against technetium pyrophosphate scintigraphy by use of our modification of the criteria suggested by Burns et al.3 6 The mean area under the CK-MB curve is also reported to provide a complementary view of the enzymatic data by use of continuous analytic methods.
We report corresponding rates of myocardial infarction (MI) by ECG criteria, primarily to provide support for the prespecified categorical end point of enzymatic MI. ECGs were obtained before surgery, on the first day after surgery, and 5 to 7 days after surgery. The digitized records were assessed by computer with the novacode algorithms, which are similar to the Minnesota Code.7 ECG MI required a preoperative to postoperative change in QS score of at least 25 in any lead; ECGs with new bundle-branch block were excluded from this analysis.
The diagnosis of low-output syndrome was determined by a majority decision from a committee of three anesthetist-intensivists, who independently reviewed the progress notes and intensive care unit flow sheets of any patient receiving inotropic medication or intra-aortic balloon pump support in the intensive care unit after surgery. Criteria for low-output syndrome were predefined: inotropic medication or intra-aortic balloon counterpulsation for ≥60 minutes in association with a thermodilution cardiac index <2.2 L · min−1 · m−2, systolic blood pressure <90 mm Hg, and a pulmonary capillary wedge pressure ≥18 mm Hg.
Continuous variables are summarized as the mean±SD or median and interquartile range in the text and tables. Between-group comparisons were performed by ANOVA. Categorical variables are summarized as the absolute frequencies or as a percentage. Between-group comparisons were performed with χ2 tests.
Logistic regression was used to analyze the effect of intermittency on the combined outcome of mortality, enzymatic MI, or low-output syndrome, controlling for other demographic predictors (left ventricular [LV] grade, presence of left main stem coronary disease, age, number of grafts placed, urgency status, and the one-way interactions of these predictors with intermittency). Predictors were limited to those that had proved to be significant in the overall trial.3 Backward stepwise logistic regression was used with a probability to remove of .20 and probability to enter of .15. Fifteen cases had a missing left ventricular (LV) grade; these were entered on the basis of surgeon, TOC, age, sex, previous cardiac surgery, preoperative MI, Canadian Cardiovascular Society (CCS) anginal class, left main stem disease, urgency, death, low-output syndrome, and enzymatic MI. A sensitivity analysis that used the best- and worst-case scenarios (ie, that assigned the 15 patients without LV grade data to grade I if they had an outcome and to grade IV otherwise or vice versa) confirmed the results found with imputation. We first determined whether interactions (as a group) were significant, and then we constructed a main effects model. Odds ratios (ORs) are reported with their 95% CI.
In a secondary analysis, we examined the robustness of the models after including variables for individual surgeons and an interaction term for surgeon×intermittency measures (ie, LTOC or PTOC). All analyses were done with stata (Computing Resource Center).
Table 1⇓ summarizes the demographic, angiographic, and operative characteristics. Table 2⇓ contains details of the administration of cardioplegia. The LTOC averaged 11.4±4.0 minutes per patient, and total TOC was 28.5±12.4 minutes, or 48.2% of the cross-clamp interval (PTOC). The patterns of practice with respect to aortic occlusion time and intermittency variables varied greatly from surgeon to surgeon (Table 2⇓). The data pertaining to LTOC (Table 3⇓) and PTOC (Table 4⇓) were divided into quartiles (LTOC, ≤9, 10 to 11, 12 to 13, and >13 minutes; PTOC, <36%, 36% to 49%, 50% to 62%, and >62%). Because of the large number of cases (n=99) with exactly the median LTOC, the groups in Table 3⇓ are of unequal size. One PTOC case was excluded owing to incomplete time data (Table 4⇓). Demographic and angiographic characteristics did not differ significantly between LTOC and PTOC quartiles.
Event rates per quartile for LTOC and PTOC are presented in Tables 3⇑ and 4⇑, respectively. Increased LTOC was harmful (>13 minutes, P=.046), whereas increased PTOC was beneficial (P=.070) by univariate statistics. MI rates as determined by QS scoring and CK-MB release reported as a continuous variable support these views.
In the multivariate analysis, there was no interaction between LTOC and nonsurgeon factors (likelihood ratio χ2, 5 df, 2.42; P=.79). Risk increased with higher LV grades (OR, 1.47; 95% CI, 1.14, 1.90; P=.003). There were nonsignificant increases in risk with left main stem disease (P=.11) and urgency (P=.19). LTOC was borderline significant (P=.070) as a risk factor for adverse outcomes: for each 1-minute increase in LTOC, the odds of an adverse outcome rose slightly (OR, 1.05; 95% CI, 1.00, 1.11). Age and number of grafts did not meet thresholds to remain in the model.
There was no significant interaction between PTOC and nonsurgeon variables (likelihood ratio χ2, 5 df, 2.00; P=.85). There was a deleterious effect with higher LV grade (OR, 1.55; 95% CI, 1.20, 2.00; P<.001) and a trend to higher risk with left main stem disease (OR, 1.60; 95% CI, 0.93, 2.74; P=.089). A borderline-significant protective effect existed for PTOC: for each 10% increase in PTOC, the OR was 0.89; 95% CI, 0.79, 1.00; P=.053. Age, urgency, and number of grafts did not meet thresholds to remain in the model.
Surgeon and surgeon×intermittency interaction terms were then added to the models (Tables 5⇓ and 6⇓). Surgeon identity was a highly significant predictor of outcome in models including LTOC and PTOC (P<.001). Both the surgeon×LTOC and surgeon×PTOC terms were of borderline significance, whereas the harmful effects of LTOC and beneficial effects of PTOC were attenuated.
The surgeon×intermittency interactions were driven almost entirely by the results of one surgeon (n=100). After exclusion of these cases, both the surgeon×LTOC (P=.79) and surgeon×PTOC (P=.18) interactions were diminished, as was the PTOC main effect (P=.76). However, longer LTOC remained marginally significant as a risk factor (P=.053; OR, 1.06; 95% CI, 1.00, 1.13).
A successful surgical outcome demands a cardioplegic method that not only maximizes myocardial preservation but also permits a superior technical result. Therefore, it is important to demonstrate that if attempts at local control fail to provide a bloodless field, warm blood cardioplegia can be interrupted safely for the period required to carefully construct a distal anastomosis under direct vision. Buckberg8 demonstrated that electromechanical arrest is associated with a 90% decrease of myocardial oxygen consumption (MV̇o2) requirements at 37°C, whereas hypothermia (10°C to 20°C) provides a further reduction of only 7.0% to 8.6%. We conjectured that short periods of normothermic ischemia would be well tolerated with warm blood cardioplegia as they are with cold blood cardioplegia, whereas extended intervals may be associated with an increased frequency of adverse outcomes. Furthermore, we anticipated that the sum of ischemic times or cumulative ischemic time would be less critical, provided that the individual ischemic times were brief and followed by adequate cardioplegic reinfusion. The differential patterns of cardioplegic interruption in the present study provided a natural experiment to evaluate the relations between intermittency and cardiac events. Our results suggest that the longest single ischemic interval is more important than the cumulative ischemic time.
For the purposes of the present investigation, patients who were crossed over from warm to cold cardioplegia and patients who received retrograde warm blood cardioplegia were excluded from the analysis to provide a group of patients who received only antegrade warm blood cardioplegia. We emphasize that the results from the primary trial favoring warm blood cardioplegia over cold blood cardioplegia were obtained with the intention-to-treat principle, ie, these exclusions were counted in the warm arm. We did not consider it worthwhile to include the warm-to-cold crossovers in the present study because then we would be evaluating the intermittent cold blood cardioplegia rather than the warm blood cardioplegia. Furthermore, patients were crossed from warm to cold owing to difficulty maintaining cardiac arrest or with excessive coronary flooding rather than because of the duration of ischemic intervals. We agree that conclusions regarding intermittency cannot be extrapolated to this population.
Our initial analysis suggested that increased intermittency according to cumulative ischemic time might in fact have a very modest protective effect on the heart. These counterintuitive protective effects of higher PTOC were weak and lost statistical significance once surgeon identity was taken into account. More important, however, the results rule out harm from increases in PTOC over a range of values typically seen in practice. As expected, prolonged LTOC (>13 minutes) was a risk factor for adverse outcomes and was not consistently abolished in models controlling for surgeon-specific effects.
In this study, cardioplegia was interrupted for ≈50% of the cross-clamp time, ranging from 22.1±8.8% to 63.8±14.5% depending on the surgeon (Table 2⇑). It is important to note that this discontinuation of cardioplegia did not occur over a single time period: the longest single ischemic interval averaged 11.4±4.0 minutes but extended from 6.8±5.4 minutes to a maximum of 15.3±5.0 minutes according to the individual surgeon (Table 2⇑). Experimental comparisons of the effects of brief ischemic periods with warm blood cardioplegia have been performed by several investigators. Landymore and associates9 addressed the time-dependent changes in lactate production and MV̇o2 in canine hearts that were associated with ischemic intervals of 1 to 10 minutes. With longer periods of ischemia, both cardiac lactate release and MV̇o2 increased in a linear fashion during cardioplegic reinfusions, to a proportionately greater degree in warm cardioplegic versus cold cardioplegic hearts. This pattern of oxygen consumption and lactate release has been noted in humans.10 These canine studies were extended in subsequent experiments in which both warm and cold cardioplegia were administered every 15 minutes.11 End-systolic and end-diastolic elastance recovered completely in both groups. Studies conducted by Chan and colleagues,12 using a similar protocol with an isovolumic preparation, provided identical conclusions regarding systolic elastance. Conversely, Ko and coworkers13 noted deterioration of systolic and diastolic functional recovery with an ejecting model after 10-minute ischemic episodes. Carrier et al14 evaluated ischemic intervals of 5, 10, 20, and 40 minutes. Increasing periods of ischemia were associated with a progressively greater reduction of intramyocardial pH and increased release of lactate, CK, and troponin T. The magnitude of change did not differ significantly between the warm and cold groups. We emphasize that the results were obtained by use of normal animal preparations.
The cumulative ischemic time using warm heart techniques for myocardial revascularization has been reported by other investigators. Cardioplegia has been interrupted for 52±23%,15 40±16%,15 39±3%,16 42±4%,17 and 29±5%10 of the aortic occlusion time, comparable to the value of 48.2±18.5% in the present study. More recently, Calafiore and associates18 report that warm blood cardioplegia was interrupted for 88±30% of the cross-clamp interval in their institution.
The tolerance of cardioplegic interruptions is likely in part dependent on the reinfusion flow rate. Yau and associates16 recommended that warm blood cardioplegia be delivered at a rate of ≥80 mL/min in coronary bypass patients according to a subgroup analysis. In the present study, ≈4.6 L blood cardioplegia (ie, 3.7 L oxygenated blood mixed with 0.9 L crystalloid additive) were administered during 62 minutes of aortic cross-clamping or approximately 75 mL/min. This represents an average flow rate of 140 mL/min when the cardioplegia was actually running. These values are comparable to other investigators’ reported results of 103 mL/min during a 54-minute cross-clamp period17 and 90 mL/min for a 72-minute interval of aortic occlusion.10 A related issue is the minimum length of time that cardioplegia should be reinfused. Evidence from experiments in a pig model suggest that 3 to 5 minutes are required.19
We excluded patients who received warm blood cardioplegia by retrograde delivery to eliminate one source of variability from our analysis. Retrograde or combined antegrade-retrograde administration of cold blood cardioplegia is widely prevalent,20 although we are uncertain whether this is true for warm cardioplegia. The addition of warm retrograde cardioplegia to antegrade cardioplegia can certainly be justified in situations in which antegrade delivery is anticipated to be incomplete.21 In experiments in a pig model performed in the setting of an acute coronary occlusion, retrograde administration was more effective than antegrade at limiting ultimate infarct size22 and restoration of systolic function.23 It should be recognized that we did not randomize patients who underwent emergency surgery directly from the catheterization laboratory in the present trial, although all other urgent patient groups are represented.3 The antegrade techniques described in our study allow for vein graft infusion of cardioplegia. Retrograde delivery may enhance protection of myocardial segments grafted with arterial conduits and be beneficial particularly for patients revascularized only with the use of arterial conduits. Retrograde administration could also be valuable when an individual anastomosis takes excessively long to complete (>13 minutes).
Alternatively, routine administration of retrograde cardioplegia may minimize the length of cardioplegic interruptions. Menasche and colleagues24 report that retrograde cardioplegia is never halted for >5 minutes. Other investigators have determined that retrograde cardioplegia flow was maintained during the cross-clamp period for 75±6% of the time.10 Despite this, lactate production and acid release from the myocardium were significantly greater with retrograde than antegrade delivery, suggesting inadequate perfusion.10 This may be due to nonnutritive flow or regional maldistribution25 as well as leakage around the self-inflating balloon into the right atrium. Note that the Emory University trial that compared cold antegrade oxygenated crystalloid cardioplegia with retrograde warm blood cardioplegia was stopped because of an excess of neurological events in the warm arm.26 While we have no explanation for this finding, both the incidence of stroke3 and findings from neuropsychological testing27 were similar in patients randomized to warm or cold cardioplegia in our randomized study.
We performed an exploratory analysis from prospectively collected data within one arm of a major randomized trial. We prespecified the parameters of the main analysis to avoid the usual problems of ersatz statistical significance related to data dredging, and we have indicated that these analyses were post hoc. We attempted to construct a clinically relevant analysis by testing PTOC and LTOC effects separately rather than simultaneously forcing both factors into the equations. The question of whether PTOC affects outcome after controlling for LTOC is of little clinical value.
A composite end point was constructed from outcome measures that favored warm cardioplegia. Power calculations conducted before the main trial recommended enrollment of 750 patients per arm. We anticipated that we would have inadequate power within the warm arm alone to test the end points separately, especially when used in multivariate models. We therefore assessed the frequency of the composite event as the primary outcome measure but provided the data for the individual parameters in addition.
Because no prestated hypothesis existed as to the tolerable limits of intermittent delivery, we determined whether any increased risk was apparent in patients across different quartiles of the LTOC and PTOC variables (a common epidemiological practice). The range of values tested for the different PTOC quartiles was broad and similar to previously reported results for coronary revascularization procedures using warm heart surgery.10 15 16 17 18 The LTOC data were more tightly clustered but within the limits of our cardioplegia technique, and they probably indicate relatively small intersurgeon differences in times for anastomotic construction. More important, these values for LTOC have been tested in animal models with precisely defined protocols.12 13 14
One of the limitations of the present study is that we did not randomize patients to different patterns of flow and interruption of cardioplegia. The intermittency variables may simply be markers for, or confounded with, other factors affecting outcome. For example, TOC may reflect to some extent the surgeon’s response to the severity of the case or the degree of complexity of the surgery, which, in turn, may directly influence outcome. Alternatively, the surgeon’s individual preferences may play a role in determining the extent of intermittency. Indeed, the individual surgeon was highly predictive of cardiac events in this study, as it was in the main trial for the combined end point of mortality and MI by ECG criteria with and without adjustment for covariates (P=.02).3 However, we could show no interactions between PTOC and LTOC and any patient or operative characteristics, apart from a borderline interaction with surgeon identity that was dependent on one operator. While any potential confounding is partially addressed by including prognostic covariates and interaction terms in our multivariate analysis, confirmatory lines of evidence from other studies are needed.
This work was supported by grants from the Medical Research Council of Canada and the Heart and Stroke Foundation of Canada. We would like to thank Lawrence Stevenson, Kay Lo, and Mara-Diana Svikis for assistance in manuscript preparation. Our greatest debt, of course, is to the Warm Heart Investigators (see list in “Appendix” of Reference 3) as well as the members of the departments of anaesthesiology and biochemistry, and the cardiovascular surgical nursing staff, whose cooperation was vital to the success of the trial.
Reprint requests to Dr Stephen Edward Fremes, Division of Cardiovascular and Thoracic Surgery, Sunnybrook Health Science Centre, 2075 Bayview Ave—H405, Toronto, Ontario M4N 3M5, Canada.
Presented in part at the 67th Scientific Sessions of the American Heart Association, Dallas, Tex, November 14-17, 1994.
- Copyright © 1995 by American Heart Association
Burns RJ, Gladstone PJ, Tremblay PC, Feindel CM, Salter DR, Lipton IH, Ogilvie RR, David TE. Myocardial infarction determined by technetium-99m pyrophosphate single-photon tomography, complicating elective coronary artery bypass grafting for angina pectoris. Am J Cardiol. 1989;63:1429-1434.
Rautaharju PM, Calhoun HP, Chaitman BR. Novacode serial ECG classification system for clinical trials and epidemiologic studies. J Electrocardiol. 1992;24(suppl):179-187.
Landymore RW, Marble AE, MacAulay MA, Fris J. Myocardial oxygen consumption and lactate production during antegrade warm blood cardioplegia. Eur J Cardiothorac Surg. 1992;6:372-376.
Chan AY, Aronov A, Hong AP, Lichtenstein SV, Abel JG. Intermittent delivery of warm blood cardioplegia. Surg Forum. 1992;43:210-213.
Ko W, Zelano J, Fahey AL, Berman K, Isom OW, Krieger KH. Ischemic tolerance of the arrested heart during warm cardioplegia. Eur J Cardiothorac Surg. 1993;4:295-299.
Yau TM, Weisel RD, Mickle DAG, Ivanov J, Mohabeer MK, Tumiati L, Carson S, Lichtenstein SV. Optimal delivery of blood cardioplegia. Circulation. 1991;84(suppl III):III-380-III-388.
Yau TM, Weisel RD, Mickle DAG, Komeda M, Ivanov J, Carson S, Mohabeer MK, Tumiati LC. Alternative techniques of cardioplegia. Circulation. 1992;86(suppl II):II-377-II-384.
Yau TM, Ikonomidis JS, Weisel RD, Mickle DAG, Ivanov J, Mohabeer MK, Tumiati L, Carson S, Liu P. Ventricular function after normothermic versus hypothermic cardioplegia. J Thorac Cardiovasc Surg. 1993;105:833-844.
Chan A, Aronov A, Hong AP, Lichtenstein SV, Abel JG. The intermittent delivery of warm blood cardioplegia. Surg Forum. 1992;42:210-213.
Misare BD, Krukenkamp IB, Lazer ZP, Levitsky S. Retrograde is superior to antegrade continuous warm blood cardioplegia for acute cardiac ischemia. Circulation. 1992;86(suppl II):II-393-II-397.
McLean RF, Wong BI, Naylor CD, Snow WG, Harrington EM, Gawel M, Fremes SE. Cardiopulmonary bypass, temperature, and central nervous system dysfunction. Circulation. 1994;90(suppl II):II-250-II-255.