journal of lnternal Medicine 1990: 227: 259-266
Cardiac involvement in bone marrow transplantation : serial changes in left ventricular size, mass and performance M. KUPARI, L. VOLIN*, A. SUOKAS, P. HEKALIt & T. RUUTU* From the Department of Cardiology, the *Department of Hematology and the tmpartment of Diagnostic Radiology. Helsinki University Central Hospital. Helsinki. Finland.
Abstract. Kupari M. Volin L. Suokas A. Hekali P, Ruutu T (Department of Cardiology, Department of Hematology and Department of Diagnostic Radiology, Helsinki University Central Hospital, Helsinki, Finland). Cardiac involvement in bone marrow transplantation : serial changes in left ventricular size, mass and performance. journal of lnternal Medicine 1990: 227: 259-266. Forty-five consecutive adult patients with haematological malignancies were studied prospectively to evaluate cardiac involvement in bone marrow transplantation (BMT). Echocardiography and measurement of systolic time intervals were performed before conditioning with cyclophosphamide (CY) (120 mg kg-I) and total body irradiation (10-12 Gy), and repeated 1 month and 1 year after BMT. The left ventricular (LV) changes at the 1-month study included increases in mass index (85.1 f 4 . 0 g m-' vs. 76.1 k 3.3 g m-2, meanf SE: P < 0.001) and in the pre-ejection period/ejection time ratio (0.46k0.01 vs. 0.36f0.01. P < 0.001). and decreases in fractional shortening (24.9*1.0% vs. 27.9+0.8%. P < 0.01) and in the peak normalized diameter lengthening rate (2.2f0.1 s-' vs. 2.6f0.1 s - I , P < 0.01). Four patients developed congestive heart failure. Twenty-four patients were alive and relapse-free 1 year after BMT. The LV measurements were then no longer different from the pre-transplant readings. Thus BMT that is preceded by conditioning with CY and total body irradiation results in increased LV mass and impaired systolic and diastolic LV function. These changes are mostly subclinical, and are also reversible if the recipient survives the initial months after transplantation.
Key words: bone marrow trarisplantation. echocardiography. heart.
Introduction The results of bone marrow transplantation (BMT) for haematological malignancies are generally favourable, but remain threatened by serious pulmonary, gastrointestinal, infectious and immunological complications [l].There are several ways in which the functioning of the heart could be compromised as well. The conditioning therapy given prior to transplantation typically includes cyclophosphamide (CY) and ionizing radiation, both of which are cardiotoxic in high doses [ 2 , 3 ] . Septic infections can be injurious to the myocardium [4],as Abbreviations: BMT = bone marrow transplantation, CY = cyclophosphamide. GVHD = graft-versus-host-disease, LV = left ventricle.
can graft-versus-host disease (GVHD) in rare cases [51. Many recipients have also previously received either doxo- or daunorubicin, both of which are cumulatively cardiotoxic [2] and could thus limit the heart's tolerance to the treatments outlined above. A number of earlier case reports [5-81, and one mainly retrospective clinical investigation [9], support the concept that effects on the heart are an important source of morbidity and mortality following BMT. One recent study [lo] found no cardiotoxicity, however, and many clinicians do not consider involvement of the heart to be an important post-transplant problem [l].We have undertaken a prospective study to evaluate the types, frequency and clinical significance of cardiac involvement in consecutive adult patients undergoing BMT for 259
260
M. KUPARI et al.
haematological malignancies. In this paper we report data on changes in left ventricular (LV) size, mass and function as recorded serially by echocardiography and systolic time intervals up to 1 year after transplantation.
Patients and methods Patients
Forty-five adult patients underwent BMT between August 1981 and January 1986 at Helsinki University Central Hospital. Each of them was entered into this study, even though complete non-invasive studies could not be performed in all cases. There were 27 men and 1 8 women aged 18-39 years; their mean age was 28.6 years. Twenty-eight individuals had acute leukaemia (20 myeloid, 8 lymphatic), 1 5 individuals had chronic myeloid leukaemia and two had Burkitt's lymphoma. Previous chemotherapy had included anthracyclines for each patient with acute leukaemia or Burkitt's lymphoma (dose range for daunorubicin doxorubicin, 116726 mg m-2 of body area : mean dose 3 70 mg m-') but for only two patients with chronic myelogenous leukaemia (doses, 118 and 3 3 3 mg m-'). All patients were free of any clinical symptoms of cardiac disease before transplantation as judged by their history, a physical examination, chest X-rays and a 12-lead electrocardiogram.
+
Treatment protocol
The essential components of the conditioning regime were intravenous CY, 6 0 mg kg-ld-' administered on two consecutive days, and total body irradiation, with 10-12 Gy given as a single dose to the first eight patients and in five to six 2-2.2 Gy daily fractions to the remainder. Shields were used to limit the lung dose to 8-10 Gy. Patients with chronic myeloid leukaemia received a single intravenous dose of daunorubicin (60 mg rn-') if they had not been treated with anthracyclines previously. The marrow allografts were donated by HLAidentical siblings ; a two-antigen mismatch was present in a single case. For suppression of GVHD, the patients were given cyclosporin for 6-1 2 months ; the first three individuals received methotrexate instead [ 111. Manifest GVHD was graded according to established criteria [ l l ] and treated with corticosteroids when necessary.
Cardiovascular follow-up
Routine clinical follow-up was provided by the haematology team and, if necessary, by a consultant cardiologist. A combined two-dimensional and Mmode echocardiographic examination and determination of the systolic time intervals were performed prior to the conditioning therapy, on discharge from hospital a median of 1 month (range, 17-120 d) post-transplantation, and at the 1-year follow-up study. Twelve-lead electrocardiograms and chest Xrays were taken at the same time, but the data obtained will be published separately, together with postmortem cardiac findings. Non-invasive techniques
The non-invasive cardiac studies were performed in the cardiovascular laboratory using a n Irex System I11 echocardiograph equipped with a 3.5 MHz phased array transducer. The echocardiographic LV examinations and the recordings for the determination of the systolic time intervals were performed by previously described methods [ 12,131. Blood pressure was measured using the cuff method. Due to technical difficulties or the patient's critical general condition we were unable to obtain high quality non-invasive data for all patients. Furthermore, 2 0 patients had died by the end of the first year following transplant, and one had a relapse of leukaemia and was therefore omitted from the late follow-up examination. Thus on completion of the study we had analysable recordings for 39 patients before BMT, for 35 individuals at the early follow-up study, and for 22 patients at the 1-year study. All recordings were analysed blindly using an x-y digitizer and computer algorithms. The echocardiographic measurements were averaged over five cardiac cycles, and the systolic time intervals were averaged over 10 cardiac cycles. The standard measurements were made according to the European recommendations [14]. The LV mass [15] was calculated using a regression formula for volume approximation [16]. The dimensions and mass of the LV were indexed by dividing by body area [17]. Fractional shortening was calculated as the difference between end-diastolic and end-systolic LV dimensions expressed as a , percentage of the end-diastolic dimension. The peak lengthening rate of the LV internal diameter was determined as described by Upton and Gibson [18], and was normalized by
CARDIAC INVOLVEMENT IN B O N E MARROW TRANSPLANTATION
dividing by the end-diastolic diameter [12]. The preejection and ejection periods of the LV were measured by the standard method, and their arithmetic ratio was taken as an index of LV function [19]. The reproducibility of these methods has previously been examined in our laboratory [12,13]. The coefficients of variation for the standard echocardiographic LV measurements and systolic time intervals were in the range 3.2-7.1% in studies repeated three to eight times over a period of 14-18 months in eight healthy subjects. The respective coefficient of variation for the peak diameter lengthening rate is 10% in our laboratory (unpublished observation). During the course of the present study we made similar non-invasive cardiac examinations in a control group of 13 men and 10 women aged 2 0 4 5 years (mean age, 28.3 years). All individuals were free of cardiovascular disease as judged by history, clinical examination and a 12-lead electrocardiogram.
261
used to assess changes in the cardiovascular variables with time; paired t-tests with the Bonferroni correction were performed where appropriate. Univariate associations between I,V measurements and selected non-cardiac or cardiac factors (age, sex, diagnosis, cumulative dose of anthracyclines per unit body area, grade of GVHD, changes in systolic blood pressure and heart rate) were tested by analysis of variance, continuity adjusted Chi-square test, and the method of least squares for correlation analysis. Factors that were found to be significant or nearly so ( P < 0.10) were subjected to multivariate analyses using multiple linear regression for continuous variables and stepwise logistic regression [20] for binary dependent variables. The normal ranges of the LV measurements were determined as the 9 5 % confidence intervals for the data in our control group ( n = 23); the upper and lower limits were set at mean f2.1 SD and mean - 2.1 SU, respectively. All tests were two-tailed; values of P < 0.05 were considered to be statistically significant. The results are expressed as mean values fSE.
Statistics Analysis of variance and covariance, followed by Duncan's multiple range test, were used to examine differences between the patient and control groups. Repeated measurement analysis of variance was
Results Cardiovascular measurements before BMT Table 1 shows the pre-transplant data for patients
Table 1. Heart rate, blood pressure and left ventricular measurements in bone marrow recipients before transplantation and in healthy subjects studied as controls (mean values k SE are shown) Group and number ( n ) of subjects studied
Measurement Heart rate (beats min-') Systolic blood pressure (mmHg) Iliastolic blood pressure (mmHg) Left ventricular end-diastolic diameter index (mm m-') end-systolic diameter index (mm mP) septa1 thickness (mm) posterior wall thickness (mm) mass index (g m-') fractional shortening (%) peak normalized diameter lengthening rate (s-I) pre-ejection period/ejection time
Acute leukaemia ( n = 30)t
Chronic myeloid leukaemia (n = IS'+
78 k 3911 115f2 69k29
70 f.49 115*4 67k3§
26.0k0.4 19.0k0.5 10.2k0.4 10.0k 0.4 76.2k2.7 27.2k1.1 2.5k0.1 0.38+0.015
26.8k0.5 18.8k0.4 9.7k0.5 1O.OkO.4 76.0 k 4.2 30.0 f1.1 2.9k0.39 0.34k0.02
'Overall P-values from analysis of variance. tPor left ventricular measurements, n = 25. $For left ventricular measurements. n = 14. S P < 0.05 compared with controls, Duncan's multiple range test. BP < 0.05 compared with patients having chronic leukaemia. Duncan's multiple range test. NS = P > 0.05.
Controls ( n = 23)
58k2 115k2 76+1 26.3 k0.4 18.6k0.4 10.1k0.3 9.1 k0.2 71.6k2.1 29.3 _+ 0.6 2.4k0.1 0.33k0.01
P*
< 0.001 NS < 0.01 NS NS NS NS NS NS < 0.05 < 0.05
262
M. K U P A R I et al.
with acute leukaemia and for patients with chronic myeloid leukaemia, and compares them with the control group : the acute leukaemia group also includes the two patients with Burkitt's lymphoma. Analysis of variance indicated significant overall differences between the three groups in heart rate, diastolic blood pressure, peak normalized lengthening rate of the LV diameter, and systolic time interval ratio. Differences in the last two variables persisted even after adjustment for different heart rates by analysis of covariance (P < 0.05 for both). Although there were no statistically significant differences in LV size, mass or function between the two patient groups, the mean values for fractional shortening, diameter lengthening rate and preejection period/ejection time showed a trend towards poorer LV performance in the acute leukaemia group (Table 1). A total of six individuals had subnormal fractional shortening ( < 24%) before BMT; five of them had acute leukaemia. Likewise, nine patients had an abnormally high systolic time interval ratio ( > 0.41), and eight of these individuals belonged to the acute leukaemia group.
LV changes after BMT Analysis of variance with repeated measurements using diagnosis as a grouping factor (acute leukaemia/chronic myeloid leukaemia) and time as a 'within' factor showed that time (i.e. BMT) had a significant effect (P < 0.01) on all other cardiovascular variables except for the LV end-systolic
diameter index. The interaction between group and time was not statistically significant for any variable, however, indicating that the cardiac response to BMT was not dependent on the haematological diagnosis. The patient groups were therefore combined for further analyses. Table 2 shows the mean changes between the pretransplant measurements noted at the 1-month and 1-year follow-up examinations. Individual data for LV mass index, fractional shortening and the systolic time interval ratio are shown in Fig. 1. The most important findings at the early follow-up study were an increase in LV mass index from 76.1 f 3.3 g m-' to 85.1 f 4 . 0 g m-' ( P < 0.001), a decrease in fractional shortening from 27.9+0.8% to 24.9f 1.0% ( P < 0.01), an increase in the systolic time interval ratio from 0.36f0.01 to 0.46k0.01 (P < 0.01), and a reduction of the peak normalized LV diameter lengthening rate from 2.6k0.1 s-' to 2.2 f O . l s-l (P < 0.01). Heart rate, systolic and diastolic blood pressure and the LV end-diastolic diameter index were also altered, indicating significant changes in the extramyocardial determinants of LV performance. All early changes in the LV measurements were statistically independent of the patient's age or sex, the diagnosis or the grade of the GVHD. The early change in fractional shortening showed a moderate inverse correlation with the concomitant change in systolic blood pressure ( r = -0.43. P < 0.05). The change in the systolic time interval ratio correlated with the pre-transplant ratio (r = -0.71, P < 0.01),
Table 2. Changes in pre-transplant left ventricular measurements between the early and late follow-up examinations after bone marrow transplantation (mean changes+SE are shown) Median time born transplantation and number In) of patients studied Measurement Heart rate (beats min-') Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Left ventricular end-diastolic diameter index (mm m-') end-systolic diameter index (mm m-*) septa1 thickness (mm) posterior wall thickness (mm) mass index (g m-') fractional shortening (%) peak normalized diameter lengthening rate (s-I) pre-ejection period/ejection time *P < 0.01, paired t-test. t P < 0.001. paired t-test.
1 month (n = 32) +14*3* +12*3* +13*3*
1 year (n
0 k 0.4 +1.5&0.3t +1.3+0.3t +9.0*2.lt -3.0k1.0' -0.4 & 0.1* + 0.10+0.02f
18)
+1k4 -4k3 +1+3
'
- 1.2 k0.4'
=
+0.1&0.6 -0.1 k0.5 -0.1 k0.3 +0.1+0.3 -0.9k2.1 +0.3 f1.0 +0.3 k 1.0 0.02 0.02
+
*
CARDIAC INVOLVEMENT IN BONE MARROW TRANSPLANTATION
150
-
Y
E
Y
: m x
I100
2
2 L
C
t e
m
2
501
,I_ CML
/
263
the change in heart rate (r = 0.47, P < 0.05), the change in LV mass index ( r = 0.55, P < 0.05) and the total cumulative dose of anthracyclines per unit body area (r = -0.44, P < 0.05). However, in multiple linear regression analysis only the concomitant change in heart rate had a significant independent effect (P < 0.05). The changes in LV mass index and peak diameter lengthening rate showed no statistical association with any of the factors tested. None of the early post-transplant changes persisted to the late follow-up study : the differences between the pre-transplant and 1-year post-transplant LV measurements were far from statistical significance (Table 2), and most of the individual 1-year data were within the normal range (Fig. 1). Post-transplant LV dysfunction
151
1
OL
I
o
-
-
-
Before BMT
-
-
d
-
+ l month
+ 1 year
Before BMT
+ 1 month
+ 1 year
Fig. 1. Individual measurements of left ventricular mass index (A), fractional shortening (B) and pre-ejection period/left ventricular ejection time (PEE'/LVET) (C). taken before bone marrow transplantation (BMT) and median times of 1 month and 1 year thereafter. Data are shown separately for patients in the acute leukaemia (AL) group and for those with chronic myeloid leukaemia (CML). Horizontal lines represent the upper and lower limits of the 95% confidence intervals for the measurements in 23 age- and sex-matched normal subjects.
Fifteen patients (47%) showed abnormally low fractional shortening at the early post-transplarh study (Fig. 1B). Compared with the 1 7 patients who had normal fractional shortening, these individuals had received a larger total dose of anthracyclines (343 & 39 mg m-2 vs. 1 4 3 f3 4 mg m-', P < 0.001) and already showed lower fractional shortening before transplantation (26.0f 1.4%vs. 31.1 +0.9%, P < 0.01). Thirteen of 19 individuals in the acute leukaemia group, but only two of 1.3 individuals with chronic myeloid leukaemia belonged to this group (P < 0.05). A stepwise logistic regression analysis revealed that after the total dose of anthracyclines had been entered into the model (improvement Chisquare 10.5, P < 0.001), neither the pre-transplant fractional shortening nor the diagnosis remained significant predictors of LV dysfunction after BMT. Four patients developed clinical and radiological signs of LV failure after BMT. Two of them had acute leukaemia and the other two were the only patients with chronic myeloid leukaemia who had received anthracyclines during their previous chemotherapy. The pre-transplant fractional shortening and the systolic time interval ratio had both been abnormal in two patients and normal in the other two. Unfortunately, only one patient could be examined non-invasively at the time of heart failure (Fig. 2). The wall thickness was markedly increased and the In several cases only one or two measurements are available instead of all three, the main reasons for missing data being technical recording difficulties and the patient's death or critical overall condition.
264
M. KUPARI et al.
Fig. 2. A l-month post-transplant echocardiographic study of a patient with acute leukaemia who developed heart failure after transplantation. The left panel is a study across the aortic (AO) root showing a dilated left atrium (LA). The centre panel shows delayed mitral valve closure (arrowheads), suggesting high end-diastolic pressure. The right panel is a left ventricular (LV) study. Note the thickness of the septum ( S ) and posterior wall (PW). the normal cavity size, normally appearing contractions and the presence of pericardial fluid (PF). Fractional shortening was 34% in this study and 32% in the pre-transplant study.
LV mass index was 6 9 % higher than before BMT; fractional shortening was maintained. Pericardial effusion was detected but there were no clinical signs of cardiac tamponade in this or any other patient. All four patients with heart failure died, and in three of them cardiac decompensation was considered to be a contributory factor.
Discussion In a prospective study of consecutive recipients of allogeneic bone marrow grafts we found a highly significant increase in LV mass index and impairment of systolic and diastolic LV performance shortly after transplantation. Nearly half (47%) of the patients developed echocardiographic evidence of systolic LV dysfunction. Although these changes were mostly subclinical and reversible, a minority ( < 10%) of patients suffered from clinically apparent cardiac decompensation. The first systematic clinical studies of heart involvement in BMT appeared in 1986 [9,10]-after a series of case reports and postmortem findings [S-81 - but yielded almost totally opposing conclusions. Cazin et al. observed, in retrospect, six fatal and 21
non-fatal cases of heart failure and/or pericarditis among 6 3 recipients of mainly autologous bone marrow grafts. They concluded that ‘cardiac toxicity is the most important limiting factor in BMT’ [9]. Baello et al., by contrast, found ‘little or no cardiotoxicity’ in 28 recipients of allogeneic marrow grafts [lo]. In the former study, most patients were given > 1 8 0 mg kg-’ of CY in the preparative regime, whereas the corresponding dose in the latter study was only 9 0 mg kg-l. In the present study, we adhered to the widely used Seattle protocol [ l l ] in the conditioning treatment, and thus all our patients received 120 mg kg-’ of CY in addition to total body irradiation. It is of interest that, like the dose of CY, the frequency of cardiac complications in our study also fell between the two earlier investigations. The key mechanism underlying the cardiotoxicity of CY is an endothelial injury that allows the exudation of fluid and blood components into the heart muscle [2 11. An oedematous myocardium has been regularly observed in fatal cases of pure experimental or clinical CY toxicity [21,22], and oedema formation also best explains the reversible post-transplant increase in LV mass that was observed in our patients. Gottdiener et al. have shown
CARDIAC INVOLVEMENT I N BONE MARROW TRANSPLANTATION
by echocardiography that the impairment of LV systolic function caused by large doses of CY ( > 180 mgkg-') becomes maximal within 2 weeks and disappears (unless fatal) by the end of the fourth week after CY treatment [22]. In our patients we observed a reduction in fractional shortening and an increase in the systolic time interval ratio -both signs of impaired LV function-a median of one month after BMT, suggesting that factors other than mere CY toxicity had a role in this change. Like CY, ionizing radiation is capable of acutely damaging the myocardial capillary endothelium [3], and Ikaheimo et al. have shown that irradiation of the mediastinum impairs LV systolic function [23]. Although the dose of radiation that the heart receives during total body irradiation is too small to cause clinically apparent cardiotoxicity on its own [3], it may well have increased and prolonged the effects of CY in our patients. GVHD and septic infections are less likely to be contributory factors, as the severity of GVHD did not appear to be related to the LV changes, and septic bacterial infections were too infrequent during the initial weeks after transplant (four patients) to contribute to the overall cardiotoxicity. Previous studies have not considered the role of altered loading conditions in the changes of LV function following BMT. In our study, the shortening of the end-diastolic diameter and the increase in blood pressure at the early post-transplant study indicate that both preload and afterload were altered in a way that tended to decrease LV ejection force. The significance of afterload is particularly significant because the change in fractional shortening was dependent on the change in systolic blood pressure : the greater the rise in blood pressure, the more marked was the decrease in fractional shortening. The increase in blood pressure was probably cyclosporin-induced [24], and was no longer present at the 1-year control study, when the drug had been discontinued in most patients. Thus altered loading conditions probably contributed significantly to the early reduction in LV systolic function following BMT. To what extent myocardial contractility was impaired remains uncertain, although the unchanged LV end-systolic dimension argues against a major negative inotropic change [2 51. The post-transplant increase in heart rate cannot be ignored either in interpretation of the data. Multiple regression analysis indicated that this change was an independent determinant of the early increase in the systolic time interval ratio. This is
265
feasible because the ratio has been shown to be directly rate-dependent when measured serially in the same subjects [26]. The post-transplant decrease in the peak lengthening rate of the LV diamter suggests impaired diastolic function : the concomitant changes in heart rate, blood pressure and fractional shortening do not explain this change [27,28]. It is highly probable that the changes in the myocardium and in the LV mass had decreased relaxation and/or passive compliance. Diastolic LV dysfunction can cause pulmonary congestion despite normal ejection performance [29], especially when combined with fluid overload, and may therefore be one significant factor in the genesis of post-transplant heart failure. Systolic LV dysfunction indicated by echocardiography or systolic time intervals was a relatively poor predictor of clinical heart problems in our study. Six patients showed fractional shortening < 24 % prior to BMT, but only two of them developed clinical heart failure after BMT. Thus the positive predictive value of pre-transplant LV dysfunction was about 30%. The corresponding value of the systolic time interval ratio was even less, since only two of nine patients with a pre-transplant ratio > 0.41 developed heart failure. The clinical usefulness of detection of asymptomatic LV systolic dysfunction after BMT was no greater, and we therefore consider that non-invasive cardiac studies are unnecessary on a routine basis in these patients.
Conclusions The present study shows that BMT preceded by treatment with CY (120 mg kg-I) and total body irradiation (10-12 GY) commonly results in impaired systolic and diastolic LV function. These changes reflect both altered loading conditions and myocardial involvement. The latter also appears as an increase in LV mass and wall thickness, probably due to oedema formation. These changes are similar in patients with acute or chronic leukaemia, but the former show more severe post-transplant LV dysfunction owing to their greater cumulative exposure to anthracyclines. The LV changes are mostly asymptomatic, and clinically apparent heart failure develops in less than 10%of recipients. The LV will return to its pre-transplant condition if the patient survives the initial months following transplantation.
266
M.K U P A R I et al.
Acknowledgements We thank the Paavo Nurmi Foundation, and the Sigrid Juselius Foundation, Helsinki, Finland, for financial support.
References 1 Champlin RE, Gale RP. The early complications of bone marrow transplantation. Semin Hematol 1 9 8 4 : 21 : 101-8. 2 Kantrowitz NE. Bristow MR. Cardiotoxicity of antitumor agents. Progr Cardiovasc Dis 1 9 8 4 ; 2 7 : 195-200. 3 Stewart JR. Fajardo LF. Radiation-induced heart disease : an update. Prog Cardiovasc Dis 1 9 8 4 : 2 7 : 173-94. 4 Parrillo JE. Cardiovascular dysfunction in septic shock : new insights into a deadly disease. Int ] Cardiol 1 9 8 5 ; 7 : 314-21. 5 Buja LM. Ferrans VJ, Craw RG. Cardiac pathologic findings in patients treated with bone marrow transplantation. Hum Pathol 1 9 7 6 ; 7 : 1 7 4 5 . 6 Santos GW. Sensebrenner L. Burke RJ et a/. Marrow transplantation in man following cyclophosphamide. Transplant Proc 1 9 7 1 ; 3 : 400-4. 7 Buckner CD, Rudolph RH, Ferer A et a/. High-dose cyclophosphamide therapy for malignant disease. Toxicity, tumor response, and the effects of stored autologous marrow. Cancer 1 9 7 2 : 2 9 : 357-65. 8 Appelbaum FR. Strauchen JA. Craw RG. Acute lethal carditis caused by high-dose combination chemotherapy. Lancet 1976 ; i: 58-62. 9 Cazin B. Gorin NC. Laporte JP et al. Cardiac complications after bone marrow transplantation. A report on a series of 63 consecutive patients. Cancer 1 9 8 6 : 57: 2061-9. 10 Baello EB. Ensberg ME, Ferguson DW et a/. Etrect of high-dose cyclophosphamide and total body irradiation on left ventricular function in adult patients with leukemia undergoing bone marrow transplantation. Cancer Treat Rep 1 9 8 6 ; 70: 1 1 87-93. 1 1 Thomas ED. Storb R. Clift RA et al. Bone marrow transplantation. N Engl] Med 1 9 7 5 : 292: 895-902. 12 Kupari M. Reproducibility of M-mode echocardiographic assessment of left ventricular function. Significance of the temporal range of measurements. Eur Heart ] 1984: 5 : 41 2-8. 1 3 Kupari M. Reproducibility of the systolic time intervals: effect of the temporal range of measurements. Cardiovasc Res 1 9 8 3 ; 17: 3 3 9 4 3 . 14 Roelandt J. Gibson DG. Recommendations for the standardization of measurements for M-mode echocardiograms. Eur Heart 1 1 9 8 0 ; 1: 375-8. 15 Feigenbaum H. Echocardiography. 4th edn. Philadelphia : Lea & Febiger. 1985: 133.
16 Teicholz LE. Kreulen T. Herman MV. Gorlin R. Problems in echocardiographic volume determination : echocardiographicangiographic correlations in the presence or absence of asynergy. Am ] Cardiol 1 9 7 6 : 3 7 : 7-12. 17 Dubois EF. Basal Metabolism in Health and Disease. Philadelphia: Lea & Febiger, 1936: 51. 1 8 Upton MT, Gibson DG. The study of left ventricular function from digitized echocardiograms. Prog Cardiovasc Res 1978 : 2 0 : 359-84. 1 9 Lewis RP. Rittgers SE, Forrester WF. Boudoulas HA. A critical review of the systolic time intervals. Circulation 1 9 7 7 : 56: 146-58. 2 0 Jennrich R, Sampson P. Frane J. Analysis of variance and covariance including repeated measures. In : Dixon WS. Brown MB, Engelman et a/.. eds. BMDP Statistical Software. Berkeley, California USA: University of California Press, 1985 : 3 59-38 7. 21 Connel TX.Berenbaum MC. Cardiac and pulmonary effects of high doses of cyclophosphamide and isophosphamide. Cancer Res 1 9 7 4 : 3 4 : 1586-91. 22 Gottdiener JS. Appelbaum FR. Ferrans V]. Deisseroth A, Ziegler J. Cardiotoxicity associated with high-dose cyclophosphamide therapy. Arch Intern Med 1981 : 141 : 753-63. 23 Ikaheimo MJ. Niemela KO, Linnaluoto MM. Jakobsson MJT. Takkunen JT. Taskinen PJ. Early cardiac changes related to radiation therapy. Am ] Cardiol 1 9 8 5 ; 56: 943-6. 24 Kahan BD. Immunosuppressive therapy with cyclosporine for cardiac transplantation. Circulation 1 9 8 7 ; 75: 40-56. 25 Marsh JD, Green LH. Wynne J. Cohn PF. Grossman W. Left ventricular end- systolic pressure dimension and stress-length relations in normal human subjects. Am J Cardiol 1979 : 4 4 : 1311-7. 26 Cokkinos CV. Heimonas ET. Demopoulos JN. Haralambakis A, Tsartsalis G. Gardikas CD. Influence of heart rate increase on uncorrected pre-ejection period/ejection time (PEP/LVET) ratio in normal individuals. Br Heart ] 1 9 7 6 : 36: 683-8. 27 Danielsen R. Nordrehaug JE, Vik-Mo H. Digitiired echocardiographic measurements of left ventricular function : reproducibility and physiological factors influencing the method. Clin Physiol 1 9 8 7 : 7: 411-21. 28 Lavine SJ, Krishnaswami V. Levinson M. Shaver JA. Etrect of heart rate alterations produced by atrial pacing on the pattern of diastolic filling in normal subjects. Am ] Cardiol 1 9 8 8 : 6 2 : 1098-102. 29 Soufer R. Wohlgelernter D. Vita NA et al. Intact systolic left ventricular function in clinical congestive heart failure. Am Cardiol 1 9 8 5 : 5 5 : 1032-6. Received 1 0 July 1989. accepted 2 8 September 1989.
Correspondence: Dr Markku Kupari. Department of Cardiology, Helsinki University Central Hospital. SF-00290 Helsinki, Finland.
Cardiac involvement in bone marrow transplantation : serial changes in left ventricular size, mass and performance M. KUPARI, L. VOLIN*, A. SUOKAS, P. HEKALIt & T. RUUTU* From the Department of Cardiology, the *Department of Hematology and the tmpartment of Diagnostic Radiology. Helsinki University Central Hospital. Helsinki. Finland.
Abstract. Kupari M. Volin L. Suokas A. Hekali P, Ruutu T (Department of Cardiology, Department of Hematology and Department of Diagnostic Radiology, Helsinki University Central Hospital, Helsinki, Finland). Cardiac involvement in bone marrow transplantation : serial changes in left ventricular size, mass and performance. journal of lnternal Medicine 1990: 227: 259-266. Forty-five consecutive adult patients with haematological malignancies were studied prospectively to evaluate cardiac involvement in bone marrow transplantation (BMT). Echocardiography and measurement of systolic time intervals were performed before conditioning with cyclophosphamide (CY) (120 mg kg-I) and total body irradiation (10-12 Gy), and repeated 1 month and 1 year after BMT. The left ventricular (LV) changes at the 1-month study included increases in mass index (85.1 f 4 . 0 g m-' vs. 76.1 k 3.3 g m-2, meanf SE: P < 0.001) and in the pre-ejection period/ejection time ratio (0.46k0.01 vs. 0.36f0.01. P < 0.001). and decreases in fractional shortening (24.9*1.0% vs. 27.9+0.8%. P < 0.01) and in the peak normalized diameter lengthening rate (2.2f0.1 s-' vs. 2.6f0.1 s - I , P < 0.01). Four patients developed congestive heart failure. Twenty-four patients were alive and relapse-free 1 year after BMT. The LV measurements were then no longer different from the pre-transplant readings. Thus BMT that is preceded by conditioning with CY and total body irradiation results in increased LV mass and impaired systolic and diastolic LV function. These changes are mostly subclinical, and are also reversible if the recipient survives the initial months after transplantation.
Key words: bone marrow trarisplantation. echocardiography. heart.
Introduction The results of bone marrow transplantation (BMT) for haematological malignancies are generally favourable, but remain threatened by serious pulmonary, gastrointestinal, infectious and immunological complications [l].There are several ways in which the functioning of the heart could be compromised as well. The conditioning therapy given prior to transplantation typically includes cyclophosphamide (CY) and ionizing radiation, both of which are cardiotoxic in high doses [ 2 , 3 ] . Septic infections can be injurious to the myocardium [4],as Abbreviations: BMT = bone marrow transplantation, CY = cyclophosphamide. GVHD = graft-versus-host-disease, LV = left ventricle.
can graft-versus-host disease (GVHD) in rare cases [51. Many recipients have also previously received either doxo- or daunorubicin, both of which are cumulatively cardiotoxic [2] and could thus limit the heart's tolerance to the treatments outlined above. A number of earlier case reports [5-81, and one mainly retrospective clinical investigation [9], support the concept that effects on the heart are an important source of morbidity and mortality following BMT. One recent study [lo] found no cardiotoxicity, however, and many clinicians do not consider involvement of the heart to be an important post-transplant problem [l].We have undertaken a prospective study to evaluate the types, frequency and clinical significance of cardiac involvement in consecutive adult patients undergoing BMT for 259
260
M. KUPARI et al.
haematological malignancies. In this paper we report data on changes in left ventricular (LV) size, mass and function as recorded serially by echocardiography and systolic time intervals up to 1 year after transplantation.
Patients and methods Patients
Forty-five adult patients underwent BMT between August 1981 and January 1986 at Helsinki University Central Hospital. Each of them was entered into this study, even though complete non-invasive studies could not be performed in all cases. There were 27 men and 1 8 women aged 18-39 years; their mean age was 28.6 years. Twenty-eight individuals had acute leukaemia (20 myeloid, 8 lymphatic), 1 5 individuals had chronic myeloid leukaemia and two had Burkitt's lymphoma. Previous chemotherapy had included anthracyclines for each patient with acute leukaemia or Burkitt's lymphoma (dose range for daunorubicin doxorubicin, 116726 mg m-2 of body area : mean dose 3 70 mg m-') but for only two patients with chronic myelogenous leukaemia (doses, 118 and 3 3 3 mg m-'). All patients were free of any clinical symptoms of cardiac disease before transplantation as judged by their history, a physical examination, chest X-rays and a 12-lead electrocardiogram.
+
Treatment protocol
The essential components of the conditioning regime were intravenous CY, 6 0 mg kg-ld-' administered on two consecutive days, and total body irradiation, with 10-12 Gy given as a single dose to the first eight patients and in five to six 2-2.2 Gy daily fractions to the remainder. Shields were used to limit the lung dose to 8-10 Gy. Patients with chronic myeloid leukaemia received a single intravenous dose of daunorubicin (60 mg rn-') if they had not been treated with anthracyclines previously. The marrow allografts were donated by HLAidentical siblings ; a two-antigen mismatch was present in a single case. For suppression of GVHD, the patients were given cyclosporin for 6-1 2 months ; the first three individuals received methotrexate instead [ 111. Manifest GVHD was graded according to established criteria [ l l ] and treated with corticosteroids when necessary.
Cardiovascular follow-up
Routine clinical follow-up was provided by the haematology team and, if necessary, by a consultant cardiologist. A combined two-dimensional and Mmode echocardiographic examination and determination of the systolic time intervals were performed prior to the conditioning therapy, on discharge from hospital a median of 1 month (range, 17-120 d) post-transplantation, and at the 1-year follow-up study. Twelve-lead electrocardiograms and chest Xrays were taken at the same time, but the data obtained will be published separately, together with postmortem cardiac findings. Non-invasive techniques
The non-invasive cardiac studies were performed in the cardiovascular laboratory using a n Irex System I11 echocardiograph equipped with a 3.5 MHz phased array transducer. The echocardiographic LV examinations and the recordings for the determination of the systolic time intervals were performed by previously described methods [ 12,131. Blood pressure was measured using the cuff method. Due to technical difficulties or the patient's critical general condition we were unable to obtain high quality non-invasive data for all patients. Furthermore, 2 0 patients had died by the end of the first year following transplant, and one had a relapse of leukaemia and was therefore omitted from the late follow-up examination. Thus on completion of the study we had analysable recordings for 39 patients before BMT, for 35 individuals at the early follow-up study, and for 22 patients at the 1-year study. All recordings were analysed blindly using an x-y digitizer and computer algorithms. The echocardiographic measurements were averaged over five cardiac cycles, and the systolic time intervals were averaged over 10 cardiac cycles. The standard measurements were made according to the European recommendations [14]. The LV mass [15] was calculated using a regression formula for volume approximation [16]. The dimensions and mass of the LV were indexed by dividing by body area [17]. Fractional shortening was calculated as the difference between end-diastolic and end-systolic LV dimensions expressed as a , percentage of the end-diastolic dimension. The peak lengthening rate of the LV internal diameter was determined as described by Upton and Gibson [18], and was normalized by
CARDIAC INVOLVEMENT IN B O N E MARROW TRANSPLANTATION
dividing by the end-diastolic diameter [12]. The preejection and ejection periods of the LV were measured by the standard method, and their arithmetic ratio was taken as an index of LV function [19]. The reproducibility of these methods has previously been examined in our laboratory [12,13]. The coefficients of variation for the standard echocardiographic LV measurements and systolic time intervals were in the range 3.2-7.1% in studies repeated three to eight times over a period of 14-18 months in eight healthy subjects. The respective coefficient of variation for the peak diameter lengthening rate is 10% in our laboratory (unpublished observation). During the course of the present study we made similar non-invasive cardiac examinations in a control group of 13 men and 10 women aged 2 0 4 5 years (mean age, 28.3 years). All individuals were free of cardiovascular disease as judged by history, clinical examination and a 12-lead electrocardiogram.
261
used to assess changes in the cardiovascular variables with time; paired t-tests with the Bonferroni correction were performed where appropriate. Univariate associations between I,V measurements and selected non-cardiac or cardiac factors (age, sex, diagnosis, cumulative dose of anthracyclines per unit body area, grade of GVHD, changes in systolic blood pressure and heart rate) were tested by analysis of variance, continuity adjusted Chi-square test, and the method of least squares for correlation analysis. Factors that were found to be significant or nearly so ( P < 0.10) were subjected to multivariate analyses using multiple linear regression for continuous variables and stepwise logistic regression [20] for binary dependent variables. The normal ranges of the LV measurements were determined as the 9 5 % confidence intervals for the data in our control group ( n = 23); the upper and lower limits were set at mean f2.1 SD and mean - 2.1 SU, respectively. All tests were two-tailed; values of P < 0.05 were considered to be statistically significant. The results are expressed as mean values fSE.
Statistics Analysis of variance and covariance, followed by Duncan's multiple range test, were used to examine differences between the patient and control groups. Repeated measurement analysis of variance was
Results Cardiovascular measurements before BMT Table 1 shows the pre-transplant data for patients
Table 1. Heart rate, blood pressure and left ventricular measurements in bone marrow recipients before transplantation and in healthy subjects studied as controls (mean values k SE are shown) Group and number ( n ) of subjects studied
Measurement Heart rate (beats min-') Systolic blood pressure (mmHg) Iliastolic blood pressure (mmHg) Left ventricular end-diastolic diameter index (mm m-') end-systolic diameter index (mm mP) septa1 thickness (mm) posterior wall thickness (mm) mass index (g m-') fractional shortening (%) peak normalized diameter lengthening rate (s-I) pre-ejection period/ejection time
Acute leukaemia ( n = 30)t
Chronic myeloid leukaemia (n = IS'+
78 k 3911 115f2 69k29
70 f.49 115*4 67k3§
26.0k0.4 19.0k0.5 10.2k0.4 10.0k 0.4 76.2k2.7 27.2k1.1 2.5k0.1 0.38+0.015
26.8k0.5 18.8k0.4 9.7k0.5 1O.OkO.4 76.0 k 4.2 30.0 f1.1 2.9k0.39 0.34k0.02
'Overall P-values from analysis of variance. tPor left ventricular measurements, n = 25. $For left ventricular measurements. n = 14. S P < 0.05 compared with controls, Duncan's multiple range test. BP < 0.05 compared with patients having chronic leukaemia. Duncan's multiple range test. NS = P > 0.05.
Controls ( n = 23)
58k2 115k2 76+1 26.3 k0.4 18.6k0.4 10.1k0.3 9.1 k0.2 71.6k2.1 29.3 _+ 0.6 2.4k0.1 0.33k0.01
P*
< 0.001 NS < 0.01 NS NS NS NS NS NS < 0.05 < 0.05
262
M. K U P A R I et al.
with acute leukaemia and for patients with chronic myeloid leukaemia, and compares them with the control group : the acute leukaemia group also includes the two patients with Burkitt's lymphoma. Analysis of variance indicated significant overall differences between the three groups in heart rate, diastolic blood pressure, peak normalized lengthening rate of the LV diameter, and systolic time interval ratio. Differences in the last two variables persisted even after adjustment for different heart rates by analysis of covariance (P < 0.05 for both). Although there were no statistically significant differences in LV size, mass or function between the two patient groups, the mean values for fractional shortening, diameter lengthening rate and preejection period/ejection time showed a trend towards poorer LV performance in the acute leukaemia group (Table 1). A total of six individuals had subnormal fractional shortening ( < 24%) before BMT; five of them had acute leukaemia. Likewise, nine patients had an abnormally high systolic time interval ratio ( > 0.41), and eight of these individuals belonged to the acute leukaemia group.
LV changes after BMT Analysis of variance with repeated measurements using diagnosis as a grouping factor (acute leukaemia/chronic myeloid leukaemia) and time as a 'within' factor showed that time (i.e. BMT) had a significant effect (P < 0.01) on all other cardiovascular variables except for the LV end-systolic
diameter index. The interaction between group and time was not statistically significant for any variable, however, indicating that the cardiac response to BMT was not dependent on the haematological diagnosis. The patient groups were therefore combined for further analyses. Table 2 shows the mean changes between the pretransplant measurements noted at the 1-month and 1-year follow-up examinations. Individual data for LV mass index, fractional shortening and the systolic time interval ratio are shown in Fig. 1. The most important findings at the early follow-up study were an increase in LV mass index from 76.1 f 3.3 g m-' to 85.1 f 4 . 0 g m-' ( P < 0.001), a decrease in fractional shortening from 27.9+0.8% to 24.9f 1.0% ( P < 0.01), an increase in the systolic time interval ratio from 0.36f0.01 to 0.46k0.01 (P < 0.01), and a reduction of the peak normalized LV diameter lengthening rate from 2.6k0.1 s-' to 2.2 f O . l s-l (P < 0.01). Heart rate, systolic and diastolic blood pressure and the LV end-diastolic diameter index were also altered, indicating significant changes in the extramyocardial determinants of LV performance. All early changes in the LV measurements were statistically independent of the patient's age or sex, the diagnosis or the grade of the GVHD. The early change in fractional shortening showed a moderate inverse correlation with the concomitant change in systolic blood pressure ( r = -0.43. P < 0.05). The change in the systolic time interval ratio correlated with the pre-transplant ratio (r = -0.71, P < 0.01),
Table 2. Changes in pre-transplant left ventricular measurements between the early and late follow-up examinations after bone marrow transplantation (mean changes+SE are shown) Median time born transplantation and number In) of patients studied Measurement Heart rate (beats min-') Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Left ventricular end-diastolic diameter index (mm m-') end-systolic diameter index (mm m-*) septa1 thickness (mm) posterior wall thickness (mm) mass index (g m-') fractional shortening (%) peak normalized diameter lengthening rate (s-I) pre-ejection period/ejection time *P < 0.01, paired t-test. t P < 0.001. paired t-test.
1 month (n = 32) +14*3* +12*3* +13*3*
1 year (n
0 k 0.4 +1.5&0.3t +1.3+0.3t +9.0*2.lt -3.0k1.0' -0.4 & 0.1* + 0.10+0.02f
18)
+1k4 -4k3 +1+3
'
- 1.2 k0.4'
=
+0.1&0.6 -0.1 k0.5 -0.1 k0.3 +0.1+0.3 -0.9k2.1 +0.3 f1.0 +0.3 k 1.0 0.02 0.02
+
*
CARDIAC INVOLVEMENT IN BONE MARROW TRANSPLANTATION
150
-
Y
E
Y
: m x
I100
2
2 L
C
t e
m
2
501
,I_ CML
/
263
the change in heart rate (r = 0.47, P < 0.05), the change in LV mass index ( r = 0.55, P < 0.05) and the total cumulative dose of anthracyclines per unit body area (r = -0.44, P < 0.05). However, in multiple linear regression analysis only the concomitant change in heart rate had a significant independent effect (P < 0.05). The changes in LV mass index and peak diameter lengthening rate showed no statistical association with any of the factors tested. None of the early post-transplant changes persisted to the late follow-up study : the differences between the pre-transplant and 1-year post-transplant LV measurements were far from statistical significance (Table 2), and most of the individual 1-year data were within the normal range (Fig. 1). Post-transplant LV dysfunction
151
1
OL
I
o
-
-
-
Before BMT
-
-
d
-
+ l month
+ 1 year
Before BMT
+ 1 month
+ 1 year
Fig. 1. Individual measurements of left ventricular mass index (A), fractional shortening (B) and pre-ejection period/left ventricular ejection time (PEE'/LVET) (C). taken before bone marrow transplantation (BMT) and median times of 1 month and 1 year thereafter. Data are shown separately for patients in the acute leukaemia (AL) group and for those with chronic myeloid leukaemia (CML). Horizontal lines represent the upper and lower limits of the 95% confidence intervals for the measurements in 23 age- and sex-matched normal subjects.
Fifteen patients (47%) showed abnormally low fractional shortening at the early post-transplarh study (Fig. 1B). Compared with the 1 7 patients who had normal fractional shortening, these individuals had received a larger total dose of anthracyclines (343 & 39 mg m-2 vs. 1 4 3 f3 4 mg m-', P < 0.001) and already showed lower fractional shortening before transplantation (26.0f 1.4%vs. 31.1 +0.9%, P < 0.01). Thirteen of 19 individuals in the acute leukaemia group, but only two of 1.3 individuals with chronic myeloid leukaemia belonged to this group (P < 0.05). A stepwise logistic regression analysis revealed that after the total dose of anthracyclines had been entered into the model (improvement Chisquare 10.5, P < 0.001), neither the pre-transplant fractional shortening nor the diagnosis remained significant predictors of LV dysfunction after BMT. Four patients developed clinical and radiological signs of LV failure after BMT. Two of them had acute leukaemia and the other two were the only patients with chronic myeloid leukaemia who had received anthracyclines during their previous chemotherapy. The pre-transplant fractional shortening and the systolic time interval ratio had both been abnormal in two patients and normal in the other two. Unfortunately, only one patient could be examined non-invasively at the time of heart failure (Fig. 2). The wall thickness was markedly increased and the In several cases only one or two measurements are available instead of all three, the main reasons for missing data being technical recording difficulties and the patient's death or critical overall condition.
264
M. KUPARI et al.
Fig. 2. A l-month post-transplant echocardiographic study of a patient with acute leukaemia who developed heart failure after transplantation. The left panel is a study across the aortic (AO) root showing a dilated left atrium (LA). The centre panel shows delayed mitral valve closure (arrowheads), suggesting high end-diastolic pressure. The right panel is a left ventricular (LV) study. Note the thickness of the septum ( S ) and posterior wall (PW). the normal cavity size, normally appearing contractions and the presence of pericardial fluid (PF). Fractional shortening was 34% in this study and 32% in the pre-transplant study.
LV mass index was 6 9 % higher than before BMT; fractional shortening was maintained. Pericardial effusion was detected but there were no clinical signs of cardiac tamponade in this or any other patient. All four patients with heart failure died, and in three of them cardiac decompensation was considered to be a contributory factor.
Discussion In a prospective study of consecutive recipients of allogeneic bone marrow grafts we found a highly significant increase in LV mass index and impairment of systolic and diastolic LV performance shortly after transplantation. Nearly half (47%) of the patients developed echocardiographic evidence of systolic LV dysfunction. Although these changes were mostly subclinical and reversible, a minority ( < 10%) of patients suffered from clinically apparent cardiac decompensation. The first systematic clinical studies of heart involvement in BMT appeared in 1986 [9,10]-after a series of case reports and postmortem findings [S-81 - but yielded almost totally opposing conclusions. Cazin et al. observed, in retrospect, six fatal and 21
non-fatal cases of heart failure and/or pericarditis among 6 3 recipients of mainly autologous bone marrow grafts. They concluded that ‘cardiac toxicity is the most important limiting factor in BMT’ [9]. Baello et al., by contrast, found ‘little or no cardiotoxicity’ in 28 recipients of allogeneic marrow grafts [lo]. In the former study, most patients were given > 1 8 0 mg kg-’ of CY in the preparative regime, whereas the corresponding dose in the latter study was only 9 0 mg kg-l. In the present study, we adhered to the widely used Seattle protocol [ l l ] in the conditioning treatment, and thus all our patients received 120 mg kg-’ of CY in addition to total body irradiation. It is of interest that, like the dose of CY, the frequency of cardiac complications in our study also fell between the two earlier investigations. The key mechanism underlying the cardiotoxicity of CY is an endothelial injury that allows the exudation of fluid and blood components into the heart muscle [2 11. An oedematous myocardium has been regularly observed in fatal cases of pure experimental or clinical CY toxicity [21,22], and oedema formation also best explains the reversible post-transplant increase in LV mass that was observed in our patients. Gottdiener et al. have shown
CARDIAC INVOLVEMENT I N BONE MARROW TRANSPLANTATION
by echocardiography that the impairment of LV systolic function caused by large doses of CY ( > 180 mgkg-') becomes maximal within 2 weeks and disappears (unless fatal) by the end of the fourth week after CY treatment [22]. In our patients we observed a reduction in fractional shortening and an increase in the systolic time interval ratio -both signs of impaired LV function-a median of one month after BMT, suggesting that factors other than mere CY toxicity had a role in this change. Like CY, ionizing radiation is capable of acutely damaging the myocardial capillary endothelium [3], and Ikaheimo et al. have shown that irradiation of the mediastinum impairs LV systolic function [23]. Although the dose of radiation that the heart receives during total body irradiation is too small to cause clinically apparent cardiotoxicity on its own [3], it may well have increased and prolonged the effects of CY in our patients. GVHD and septic infections are less likely to be contributory factors, as the severity of GVHD did not appear to be related to the LV changes, and septic bacterial infections were too infrequent during the initial weeks after transplant (four patients) to contribute to the overall cardiotoxicity. Previous studies have not considered the role of altered loading conditions in the changes of LV function following BMT. In our study, the shortening of the end-diastolic diameter and the increase in blood pressure at the early post-transplant study indicate that both preload and afterload were altered in a way that tended to decrease LV ejection force. The significance of afterload is particularly significant because the change in fractional shortening was dependent on the change in systolic blood pressure : the greater the rise in blood pressure, the more marked was the decrease in fractional shortening. The increase in blood pressure was probably cyclosporin-induced [24], and was no longer present at the 1-year control study, when the drug had been discontinued in most patients. Thus altered loading conditions probably contributed significantly to the early reduction in LV systolic function following BMT. To what extent myocardial contractility was impaired remains uncertain, although the unchanged LV end-systolic dimension argues against a major negative inotropic change [2 51. The post-transplant increase in heart rate cannot be ignored either in interpretation of the data. Multiple regression analysis indicated that this change was an independent determinant of the early increase in the systolic time interval ratio. This is
265
feasible because the ratio has been shown to be directly rate-dependent when measured serially in the same subjects [26]. The post-transplant decrease in the peak lengthening rate of the LV diamter suggests impaired diastolic function : the concomitant changes in heart rate, blood pressure and fractional shortening do not explain this change [27,28]. It is highly probable that the changes in the myocardium and in the LV mass had decreased relaxation and/or passive compliance. Diastolic LV dysfunction can cause pulmonary congestion despite normal ejection performance [29], especially when combined with fluid overload, and may therefore be one significant factor in the genesis of post-transplant heart failure. Systolic LV dysfunction indicated by echocardiography or systolic time intervals was a relatively poor predictor of clinical heart problems in our study. Six patients showed fractional shortening < 24 % prior to BMT, but only two of them developed clinical heart failure after BMT. Thus the positive predictive value of pre-transplant LV dysfunction was about 30%. The corresponding value of the systolic time interval ratio was even less, since only two of nine patients with a pre-transplant ratio > 0.41 developed heart failure. The clinical usefulness of detection of asymptomatic LV systolic dysfunction after BMT was no greater, and we therefore consider that non-invasive cardiac studies are unnecessary on a routine basis in these patients.
Conclusions The present study shows that BMT preceded by treatment with CY (120 mg kg-I) and total body irradiation (10-12 GY) commonly results in impaired systolic and diastolic LV function. These changes reflect both altered loading conditions and myocardial involvement. The latter also appears as an increase in LV mass and wall thickness, probably due to oedema formation. These changes are similar in patients with acute or chronic leukaemia, but the former show more severe post-transplant LV dysfunction owing to their greater cumulative exposure to anthracyclines. The LV changes are mostly asymptomatic, and clinically apparent heart failure develops in less than 10%of recipients. The LV will return to its pre-transplant condition if the patient survives the initial months following transplantation.
266
M.K U P A R I et al.
Acknowledgements We thank the Paavo Nurmi Foundation, and the Sigrid Juselius Foundation, Helsinki, Finland, for financial support.
References 1 Champlin RE, Gale RP. The early complications of bone marrow transplantation. Semin Hematol 1 9 8 4 : 21 : 101-8. 2 Kantrowitz NE. Bristow MR. Cardiotoxicity of antitumor agents. Progr Cardiovasc Dis 1 9 8 4 ; 2 7 : 195-200. 3 Stewart JR. Fajardo LF. Radiation-induced heart disease : an update. Prog Cardiovasc Dis 1 9 8 4 : 2 7 : 173-94. 4 Parrillo JE. Cardiovascular dysfunction in septic shock : new insights into a deadly disease. Int ] Cardiol 1 9 8 5 ; 7 : 314-21. 5 Buja LM. Ferrans VJ, Craw RG. Cardiac pathologic findings in patients treated with bone marrow transplantation. Hum Pathol 1 9 7 6 ; 7 : 1 7 4 5 . 6 Santos GW. Sensebrenner L. Burke RJ et a/. Marrow transplantation in man following cyclophosphamide. Transplant Proc 1 9 7 1 ; 3 : 400-4. 7 Buckner CD, Rudolph RH, Ferer A et a/. High-dose cyclophosphamide therapy for malignant disease. Toxicity, tumor response, and the effects of stored autologous marrow. Cancer 1 9 7 2 : 2 9 : 357-65. 8 Appelbaum FR. Strauchen JA. Craw RG. Acute lethal carditis caused by high-dose combination chemotherapy. Lancet 1976 ; i: 58-62. 9 Cazin B. Gorin NC. Laporte JP et al. Cardiac complications after bone marrow transplantation. A report on a series of 63 consecutive patients. Cancer 1 9 8 6 : 57: 2061-9. 10 Baello EB. Ensberg ME, Ferguson DW et a/. Etrect of high-dose cyclophosphamide and total body irradiation on left ventricular function in adult patients with leukemia undergoing bone marrow transplantation. Cancer Treat Rep 1 9 8 6 ; 70: 1 1 87-93. 1 1 Thomas ED. Storb R. Clift RA et al. Bone marrow transplantation. N Engl] Med 1 9 7 5 : 292: 895-902. 12 Kupari M. Reproducibility of M-mode echocardiographic assessment of left ventricular function. Significance of the temporal range of measurements. Eur Heart ] 1984: 5 : 41 2-8. 1 3 Kupari M. Reproducibility of the systolic time intervals: effect of the temporal range of measurements. Cardiovasc Res 1 9 8 3 ; 17: 3 3 9 4 3 . 14 Roelandt J. Gibson DG. Recommendations for the standardization of measurements for M-mode echocardiograms. Eur Heart 1 1 9 8 0 ; 1: 375-8. 15 Feigenbaum H. Echocardiography. 4th edn. Philadelphia : Lea & Febiger. 1985: 133.
16 Teicholz LE. Kreulen T. Herman MV. Gorlin R. Problems in echocardiographic volume determination : echocardiographicangiographic correlations in the presence or absence of asynergy. Am ] Cardiol 1 9 7 6 : 3 7 : 7-12. 17 Dubois EF. Basal Metabolism in Health and Disease. Philadelphia: Lea & Febiger, 1936: 51. 1 8 Upton MT, Gibson DG. The study of left ventricular function from digitized echocardiograms. Prog Cardiovasc Res 1978 : 2 0 : 359-84. 1 9 Lewis RP. Rittgers SE, Forrester WF. Boudoulas HA. A critical review of the systolic time intervals. Circulation 1 9 7 7 : 56: 146-58. 2 0 Jennrich R, Sampson P. Frane J. Analysis of variance and covariance including repeated measures. In : Dixon WS. Brown MB, Engelman et a/.. eds. BMDP Statistical Software. Berkeley, California USA: University of California Press, 1985 : 3 59-38 7. 21 Connel TX.Berenbaum MC. Cardiac and pulmonary effects of high doses of cyclophosphamide and isophosphamide. Cancer Res 1 9 7 4 : 3 4 : 1586-91. 22 Gottdiener JS. Appelbaum FR. Ferrans V]. Deisseroth A, Ziegler J. Cardiotoxicity associated with high-dose cyclophosphamide therapy. Arch Intern Med 1981 : 141 : 753-63. 23 Ikaheimo MJ. Niemela KO, Linnaluoto MM. Jakobsson MJT. Takkunen JT. Taskinen PJ. Early cardiac changes related to radiation therapy. Am ] Cardiol 1 9 8 5 ; 56: 943-6. 24 Kahan BD. Immunosuppressive therapy with cyclosporine for cardiac transplantation. Circulation 1 9 8 7 ; 75: 40-56. 25 Marsh JD, Green LH. Wynne J. Cohn PF. Grossman W. Left ventricular end- systolic pressure dimension and stress-length relations in normal human subjects. Am J Cardiol 1979 : 4 4 : 1311-7. 26 Cokkinos CV. Heimonas ET. Demopoulos JN. Haralambakis A, Tsartsalis G. Gardikas CD. Influence of heart rate increase on uncorrected pre-ejection period/ejection time (PEP/LVET) ratio in normal individuals. Br Heart ] 1 9 7 6 : 36: 683-8. 27 Danielsen R. Nordrehaug JE, Vik-Mo H. Digitiired echocardiographic measurements of left ventricular function : reproducibility and physiological factors influencing the method. Clin Physiol 1 9 8 7 : 7: 411-21. 28 Lavine SJ, Krishnaswami V. Levinson M. Shaver JA. Etrect of heart rate alterations produced by atrial pacing on the pattern of diastolic filling in normal subjects. Am ] Cardiol 1 9 8 8 : 6 2 : 1098-102. 29 Soufer R. Wohlgelernter D. Vita NA et al. Intact systolic left ventricular function in clinical congestive heart failure. Am Cardiol 1 9 8 5 : 5 5 : 1032-6. Received 1 0 July 1989. accepted 2 8 September 1989.
Correspondence: Dr Markku Kupari. Department of Cardiology, Helsinki University Central Hospital. SF-00290 Helsinki, Finland.
