Author + information
- Received April 8, 1998
- Revision received June 15, 1999
- Accepted August 12, 1999
- Published online November 15, 1999.
- ↵*Reprint requests and correspondence: I. Östman-Smith, Department of Pediatrics, John Radcliffe Hospital, Headington, Oxford OX3 9DU United Kingdom
The study analyzed factors, including treatment, affecting disease-related death in patients with hypertrophic cardiomyopathy (HCM) presenting in childhood.
Previous smaller studies suggest that mortality is higher in patients with HCM presenting in childhood compared with presentation in adulthood, but these studies have all originated from selected patient populations in tertiary referral centers, and reported no significant protection by treatment.
Retrospective comparisons of mortality were done in total cohort of patients presenting to three regional centers of pediatric cardiology. There were 66 patients (25 with Noonan’s syndrome) with HCM presenting at age <19 years; mean follow-up was 12.0 years.
Among risk factors for death were congestive heart failure (p = 0.008), large electrocardiogram voltages (Sokolow-Lyon index p = 0.0003), and degree of septal (p = 0.004) and left ventricular (p = 0.028) hypertrophy expressed as percent of 95th centile value. The only treatment that significantly reduced the risk of death on multifactorial analysis of variance was high-dose beta-adrenoceptor antagonist therapy (propranolol 5 to 23 mg/kg/day or equivalent; p = 0.0001). Nineteen out of 40 patients managed conventionally (no treatment, 0.8 to 4 mg/kg of propranolol, or verapamil) died, median survival 15.8 years, with no deaths among 26 patients on high-dose beta-blockers (p = 0.0004); survival proportions at 10 years were 0.65 (95% confidence interval 0.49–0.80) and 1.0, respectively (p = 0.0015). Survival time analysis shows better survival in the high-dose beta-blocker group compared with the “no specific therapy” group (p = 0.0009) and with the conventional-dose beta-blocker group (p = 0.002). Hazard ratio analysis suggests that high-dose beta-blocker therapy produces a 5–10- fold reduction in the risk of disease-related death.
High-dose beta-blocker therapy improves survival in childhood HCM.
Hypertrophic cardiomyopathy (HCM) presenting in childhood has higher mortality (annual rate 4.6% to 5.8%) (1–3)than in adult life (1% to 4%) (4,5). It has, however, been suggested that high mortality figures are due to referral bias of severe cases to institutions with research interests in HCM (6). Observational studies in adults have suggested possible benefits from treatment with verapamil (4), amiodarone (7), or high-dose propranolol (8–10), and symptomatic infants and children obtain considerable benefit from standard and high-dose propranolol treatment (11,12). The aim of this study was to compare the impact of different treatment strategies on the survival of patients with HCM, with and without Noonan’s syndrome, presenting in childhood to three regional centers with a geographical basis for referral.
All patients with a diagnosis of HCM made before 19 years of age, and attending the Departments of Pediatric Cardiology at the University Hospital (Lund), the Academic Hospital (Uppsala), and the John Radcliffe Hospital (Oxford), over the last 30 years, were included in the study. In surviving patients at least one year of follow-up was required for inclusion. All three hospitals were regional centers with a geographical basis for patient referrals, and the senior clinicians in each department were the same throughout the study period, giving a consistency in approach to treatment. Original patient records and echocardiographic data were re-examined, taking great care to exclude all patients where the HCM was secondary to maternal gestational diabetes, mitochondrial disorders, myopathies or Friedreich’s ataxia, or secondary to medical therapy. A total of 66 patients with true primary HCM were identified. The diagnosis of Noonan’s syndrome was based on short stature and typical dysmorphic features. Follow-up duration was up to 32.4 years (mean 12 years), and follow-up information was obtained on all patients that had moved. No patients were lost to follow-up. Autopsy information and histological confirmation of the diagnosis were both available on all subjects that died, except one child with Noonan’s syndrome and classical severe HCM on echocardiography.
The treatment regimes identified were: 1) No medical treatment at all, or diuretic therapy only, the “No Specific Therapy” (NST) group (n = 20); 2) Beta-adrenoceptor antagonist therapy, propranolol in 89%, 1 to 23 mg/kg/day (n = 41); 3) Calcium-channel blockers (mostly verapamil 2.2 to 8.5 mg/kg/day; n = 6). Fifteen patients following treatment regimes 2 and 3 had additional anti-arrhythmic therapy (quinidine, amiodarone, sotalol or disopyramide).
Original electrocardiogram (ECG) tracings and M-mode echocardiographic tracings were remeasured using standardized criteria. Sokolow-Lyon index (S in lead V1+ R in lead V5or lead V6, whichever is largest) (13)and the sum totals of R+S voltages in limb leads were calculated. To judge relative severity of cardiac hypertrophy, two types of ratios were employed. First, the ratios of septal and posterior left ventricular (LV) wall thickness were divided by the diastolic LV cavity diameter (septum-to-cavity ratio and LV wall-to-cavity ratio). Second, the observed values of septal and posterior wall thickness were expressed as percent of the respective predicted 95th centile values for age. Data on normal wall thickness values related to body surface area were used for comparison with infants up to six months of age (14), whereas for older children the 95th centile prediction limits for age were calculated from 200 normal children age 0 to 18 years in Oxford. These normal data corresponded to those published by Feigenbaum (15)for selected weight categories and at 18 years predicted values equal to the upper limit of normal for adults. The 95th centile prediction limit was calculated by the equation 0.625 + (age in years × 0.0269) for septal thickness and by 0.565 + (age in years × 0.030) for posterior LV wall thickness.
Statistical analysis was carried out using commercial software (Statgraphics Plus, PRISM and CIA). Kaplan-Meier survival curves with log-rank ttest, and confidence limits for survival proportions and hazard ratios, were all calculated. The Fisher exact test was used for categorical data analysis. The Mann-Whitney Utest for unpaired data was used for intergroup comparisons of variables. Risk factors for death were analyzed by multivariate correlation analysis (Statgraphics Plus). Multifactor analysis of variance (ANOVA) was used to test that the effect of treatment was independent from chance influence of risk factor distribution (Statgraphics Plus). Response variables were death or survival, coded as integers, and treatment modality was the first factor. Only up to three factors could be analyzed at a time, because of degrees of freedom available, and thus the only treatment to show a significant effect was analyzed sequentially together with all identified risk factors. A significance level of <0.05 of the interaction between treatment and another factor would be evidence that the treatment effect was not independent.
The patient groups in the three centers were very similar, with no significant difference in the age of diagnosis, clinical characteristics, proportion of cases with Noonan’s syndrome, or length of follow-up (see Table 1). Accordingly, the patients were combined and analyzed as a single cohort in terms of analysis of mortality and risk factors, although HCM associated with Noonan’s syndrome was also analyzed separately and compared with HCM not associated with Noonan’s syndrome.
There were 66 patients with 789 patient years and 19 deaths: 10 deaths were sudden, 4 were due to severe congestive heart failure (CCF) in infants, 3 to sequelae of late CCF in older patients, and 2 were related to surgical myectomy. However, there were suggestive differences between the annual mortality rates among the dominant treatment regimes. The NST group had an annual mortality of 6.6% (n = 20), with 3.5% annual mortality in the totally asymptomatic subjects (n = 14). The beta-adrenoceptor antagonist group (n = 41) had an annual mortality of 2.1%, and the calcium-channel blocker group (n = 6) had the highest annual mortality, 10.5%.
Effect of treatment on mortality
Correlation analysis to identify treatment factors influencing survival in the combined groups showed a significant positive correlation between propranolol dose and survival (p = 0.002), whereas calcium channel-blocker therapy showed no correlation with either death or survival (p = 0.26; 0.63). As the calcium-channel blocker group was very small, no further statistical comparisons were made with this group. Figure 1illustrates the beta-adrenoceptor blocking dose in propranolol-equivalents in survivors and nonsurvivors treated with beta-blockers, and it is notable that all the deaths had occurred with doses in the low range (mean dose [standard error, SE] in nonsurvivors = 2.0 [0.2] mg/kg/day as compared with 7.8 [1.0] mg/kg/day in survivors; p = 0.015 Mann-Whitney Utest). As previously published studies had found no protection from conventional doses of propranolol, the beta-blocker group was subdivided into conventional-dose therapy (CDβB, n = 18 patients; 0.8 to 4 mg/kg/day), and high-dose therapy (HDβB, n = 26; 5 to 23 mg/kg/day of propranolol, or equivalent doses of metoprolol or atenolol in 3/26). There was a highly significant correlation between HDβB and survival (p = 0.0002; no deaths among 26 patients on this treatment). However, CDβB showed no significant correlation with survival, and this group had an annual mortality of 4.3%.
Multifactor analysis of variance (ANOVA) confirmed that out of all treatment regimes only HDβB significantly reduced disease-related death (p = 0.0001). A significant proportion (10/26) of the patients in the HDβB group had additional therapy with disopyramide, and accordingly the group was further divided in HDβB as monotherapy, and in HDβB combined with disopyramide. In the CDβB group, 3/18 patients had additional anti-arrhythmic therapy (sotalol or amiodarone). Table 2shows the clinical characteristics of all the cases combined, and of the various identified treatment subgroups. Four patients had treatment years on two regimes included in analysis as they were alive on the second regime: one with 5 years in the NST group and 2.2 years on CDβB; one with 4.0 years on CDβB and 8.2 years on HDβB and disopyramide; one with 16.0 years on CDβB and 4.2 years on HDβB and disopyramide; one with 2.8 years HDβB followed by 8.7 years of CDβB. Apart from less severe degree of cardiac hypertrophy in the largely asymptomatic NST group and no patients with a family history of sudden death in this group, the treatment groups are similar in all potential risk factors, except that the combined HDβB plus disopyramide group tended toward more severe hypertrophy.
Kaplan-Meier survival-time analysis showed that the survival proportion of patients in the monotherapy HDβB group was better than that of the NST group (log-rank test p = 0.006), and of that of the monotherapy CDβB group (p = 0.03; Fig. 2A). There was no significant difference between the monotherapy CDβB group and the NST group (Fig. 2A). The group with combined disopyramide and HDβB (n = 10; patients indicated by diamonds in Fig. 2B) also had a significantly better survival on log-rank test than did the NST group (p = 0.02) and the monotherapy CDβB group (p = 0.03). Comparing the total CDβB group including patients with anti-arrhythmic therapy with the total HDβB group (with and without disopyramide), their survival curves also show a highly significant difference in survival (p = 0.004); likewise, the difference in survival between the total HDβB group and the NST group is highly significant (p = 0.0009); see Figure 2B.
To quantify the treatment effect of high-dose beta-blockers, the hazard ratios for death in the major treatment groups are compared with the hazard of death in the NST group set as 1.0 (as the risk approximating to natural history) in Table 3. Only the HDβB groups, with or without disopyramide, show a significant reduction in the hazard of death, with the hazard ratios suggesting a 5–10-fold reduction in risk.
As there were no significant differences in the survival proportions among the NST, CDβB, and calcium-channel blocker groups, these three treatment groups were pooled as “conventional management” for a Kaplan-Meier plot illustrating total survival from the time of diagnosis (Fig. 3). The advantage of this plot was that it allowed for the inclusion of all treatment years in patients changing between different treatment regimes, and of those dying on their second regime, thus giving a truer picture of median survival after diagnosis and of the annual mortality. With “conventional management” there were 19 deaths among 40 patients (44%), annual mortality 4.0%, as against no deaths among the 26 patients on HDβB therapy (p = 0.0004 on the Fisher exact test), and the difference in survival is highly significant on the log-rank test (p = 0.0015). The survival proportion 10 years after the diagnosis was estimated at 0.65 (95% CI [confidence interval] 0.49 to 0.80) in the “conventional management” group, and the median survival from diagnosis was 15.8 years.
In the “conventional management” group, 10/19 deaths were sudden. The median age at sudden death was 13.3 years, and 6/10 sudden deaths occurred in the age range 9.8 to 13.5 years. However, there were no deaths in HDβB group (p = 0.006; Fisher exact test). Multiple ANOVA confirms that HDβB significantly reduces the risk of sudden death (p = 0.0018).
To ascertain that the treatment groups were comparable in their risk factors, we analyzed the clinical features correlating with death in our cohort of patients. As HDβB therapy had a significantly protective effect, patients receiving this therapy were removed from the entire group for statistical analysis of the risk factors for disease-related death. In the remaining patients (n = 40) multivariate correlation analysis showed that the presence of heart failure was positively correlated with the occurrence of death (p = 0.002). The presence of ECG evidence of hypertrophy in the form of a large Sokolow-Lyon index (p = 0.0003) and large sum of RS-voltages in limb leads (p = 0.0008) also showed a highly significant correlation with death. Echocardiographic indices of cardiac hypertrophy at diagnosis also showed positive correlation with death, both for septum-to-cavity ratio (p = 0.021), diastolic LV wall-to-cavity ratio (p = 0.043) and absolute septal and LV wall thickness normalized for age by being expressed as percentage of 95th centile value (see Methods; p = 0.004 and p = 0.028). Multiple ANOVA on the total cohort of 66 patients shows no significant interaction between any of the identified risk factors at presentation and the treatment effect of HDβB.
Influence of Noonan’s syndrome
To determine whether the effect of HDβB was affected by the genetic mechanism underlying the HCM, we analyzed separately the effect of HDβB in the 25 patients with Noonan’s syndrome and HCM, and in HCM not associated with Noonan’s syndrome. In these subgroups there were no deaths among eight Noonan patients on HDβB, but 10/17 of the Noonan HCM patients on other regimes died (p = 0.008 on the Fisher exact test). Survival time analysis with log-rank test confirms that HDβB confers a better survival than “conventional management” both in HCM associated with Noonan’s syndrome (p = 0.03; see Fig. 4A), and with non-Noonan childhood HCM (p = 0.02; see Fig. 4B).
Among patients not on HDβB, sudden unexpected deaths occurred in 4/17 patients with Noonan’s syndrome (annual rate 2.3%), and in 6/21 patients with familial or sporadic HCM (annual rate 2.3%).
Six patients proceeded to myectomy for reduction of LV outflow tract obstruction, one from the NST group, three from the conventional-dose beta-blocker group, and two from the calcium-channel-blocker group. There was one reoperation, two perioperative deaths, and two late deaths (annual mortality after the first operation 11.4%). The two medium-term survivors have recurrence of outflow-obstruction or angina; thus, there is no patient with a satisfactory medium-term result.
Any beneficial effect of disopyramide on survival cannot be separated from that of HDβB as all 10 patients receiving disopyramide are also in the HDβB group. The indications for disopyramide were ventricular tachycardia, paroxysmal atrial fibrillation with Wolff-Parkinson-White syndrome, and in children ventricular ectopics (Lown grade 2–3) or dynamic outflow-obstruction persisting despite adequate beta-adrenoceptor blockade. These patients remain in sinus rhythm with satisfactory arrhythmia control and with no deaths over 85 patient years in this high-risk group.
Five patients received amiodarone therapy at some stage, in doses of 100 (4 patients) to 200 mg/day. Three patients had amiodarone discontinued because of side effects; and one discontinued amiodarone as she was symptomatically worse. There was one death (sudden) in 14 patient years, and no patient did well over the long term.
Side effects of drugs
Amiodarone had the highest proportion of serious side effects (pulmonary, hepatic and thyroid), but verapamil was also associated with a notably high proportion of patients developing heart failure and progressing to a dilated end-stage (3/6). The high-dose beta-blocker therapy was tolerated in all patients, although in older asymptomatic patients the dose has to be increased slowly over 6 to 12 months to avoid undue fatigue on exertion. Three patients have changed totally or partly from propranolol to metoprolol or atenolol due to bronchospasm (1 patient) or nightmares (2 patients). Two patients have had an episode of hypoglycemia (with no sequelae) after more than 18 h fasting, and parents should be advised to avoid prolonged periods without food.
Patient acceptability of high-dose beta-blocker
Patients in the higher beta-blocker dose range had their doses increased from a starting dose of 1.5 to 3 mg/kg/day of propranolol to a minimum of 6 mg/kg/day. Further dose increases were determined by therapeutic response, and 24-h Holter monitoring was used to indicate that a profound beta-receptor blockade had been achieved (see Fig. 5). This usually requires serum concentrations of propranolol between 200 and 900 μg/liter. When changing from plain to slow-release propranolol a dose increase may be needed to compensate for less complete absorption, particularly in young children. Patients adapt well to profound beta-blockade and are able to manage schooling with good academic results and full-time employment. Patients with asthma have tolerated treatment with high-dose metoprolol well, as long as prophylactic inhalers are used regularly. Three pregnancies have proceeded uneventfully, with a mother on 7.5 mg/kg of metoprolol, producing two healthy infants with birth weights on the 3rd centile, and a mother on 3.8 mg/kg of atenolol, 8.2 mg/kg of propranolol and 10.7 mg/kg of disopyramide producing an infant on the 50th centile with no neonatal problems. Any changeover from nonselective to selective beta-blockers needs to be carried out gradually with repeated 24-h ECG monitoring, as we have seen re-emergence of previously well-controlled ventricular arrhythmias when attempting to change from propranolol to atenolol, and in one case had to settle for combined treatment of atenolol and a reduced dose of propranolol. Maximum exercise capability was curtailed somewhat in patients with only mild hypertrophy at the start of treatment, but generally improved in patients with severe hypertrophy at the start of high-dose therapy. Only one (asymptomatic) patient elected to discontinue high-dose therapy because of reduced exercise-tolerance.
Mortality in a cohort study
The relative rarity of HCM means that there is no report in the literature of any prospective, randomized study of the long-term outcome of any medical therapy for HCM. The even greater rarity of HCM presenting in childhood makes it necessary to use retrospective data to achieve sufficient patient years for an initial comparison of different medical therapies. The strength of the current study is that it assesses the outcome of a total cohort of patients arising from regional centers with a geographical basis for referral, therefore avoiding recruitment bias and avoiding an over-representation of severe cases as might be seen in tertiary centers with a research interest in HCM. It is therefore not surprising that the total mortality observed in the conventionally managed group (4.0%) is slightly lower than figures previously reported from specialized supraregional centers in the United Kingdom and the U.S. (annual mortalities of 4.6% to 5.9%) (1–3). Even with the lower rate of mortality in this study, the median survival with conventional management is only 15.8 years, and 10/19 deaths were sudden, with the majority occurring in asymptomatic 9 to 13-year-olds.
Role of progressive hypertrophy in childhood HCM
Hypertrophic cardiomyopathy is a genetically heterogeneous condition, and over 70 mutations, all encoding contractile proteins, have now been described (16). It has been suggested that the cardiac hypertrophy seen in HCM is a compensatory phenomenon rather than being directly caused by the mutations (16). Most adults presenting with HCM were asymptomatic as children, but cardiac hypertrophy due to HCM can be rapidly progressive in childhood, particularly during adolescence (17). Previous studies have looked for risk factors for sudden death only, and in HCM patients of all ages failed to demonstrate a significant correlation with absolute wall thickness (18). The present study includes all disease-related deaths, and demonstrates that in our cohort of HCM patients presenting in childhood the risk of all disease-related death, and of sudden death, increases with increasing cardiac hypertrophy when it is expressed relative either to patient age or to cavity size. The inference is therefore that medical therapy aimed at reducing progression of cardiac hypertrophy might be of benefit.
Effect of treatment on mortality in childhood HCM
The only medical therapy that significantly reduced the risk of death on multivariate ANOVA was HDβB therapy. No deaths occurred among 26 patients with 199 patient years of observation on high-dose beta-blocker treatment, including 113 patient years on monotherapy with beta-blockers. On survival time analysis, both the group with monotherapy and the group with high-dose beta-blocker and disopyramide had a highly significant survival advantage compared with both the NST group, whose annual mortality was 6.6%, and the conventional-dose beta-blocker group, whose annual mortality was 4.3%. This result occurred despite the high-dose beta-blocker group being extremely well matched in risk factors with the conventional-dose group, and having a higher incidence of risk factors than the NST group. The size of this treatment effect appears large, with hazard ratios suggesting protection in the order of a 5 to 10-fold reduction in mortality. The treatment effect appears independent of the genetic mechanism underlying the HCM, as it is present both in HCM without and with Noonan’s syndrome (Fig. 4). Treatment with conventional doses of beta-blocker, particularly in the lower range of 2 to 3 mg/kg/day, did not protect against the occurrence of sudden death, in agreement with previous studies (18).
Potential mechanism of the effect of beta-adrenoceptor antagonists
Normal myocardial cells show both age-related growth and compensatory hypertrophy. Animal experimental evidence suggests that increased cardiac sympathetic nervous activity is the most important trigger of compensatory cardiac hypertrophy (19). Furthermore, animal experimental work has shown that, although high-dose beta-adrenoceptor blockade does not reduce age-related cardiac growth (20), it greatly reduces compensatory cardiac hypertrophy through a mechanism independent of effect on cardiac workload (20). In vivo this effect is exerted largely via beta1-adrenoceptors (21). Increased activity of sympathetic nerves in the heart is present in patients with HCM (22)and may well be a compensatory reflex response to the small LV cavity (23)and to the impaired diastolic filling in HCM (24). Propranolol treatment improves diastolic function in HCM (25,26), but only large doses (480 mg/day in adults) return iso-volumic relaxation time to normal (26). In the group studied here a significant reduction in cardiac hypertrophy in the group on high-dose beta-blocker therapy was observed (12). As the heart is growing rapidly in childhood it may represent a favorable window of opportunity for pharmacological intervention.
There are earlier studies on adults with HCM that also report very low (0% to 0.6%) annual mortality rates with high-dose propranolol treatment (9,10), but neither of these studies had control groups. It is known that conventional and medium-sized doses of propranolol do not abolish ventricular tachycardia in HCM (27,28), but it has been speculated that propranolol might prevent ventricular tachycardia degenerating into ventricular fibrillation (28), similar to its postulated protective effect in ischemic heart disease (29), and in this respect propranolol is probably better than beta1-selective drugs.
A substantial proportion of the HDβB group had additional anti-arrhythmic treatment with disopyramide, a drug that has beneficial hemodynamic effects in HCM (30,31)with actions that are additive to the effects of propranolol (32). This combination may have contributed to the favorable outcome in the HDβB group combined with disopyramide, which had particularly severe hypertrophy, but is not the sole cause of it, as disopyramide alone does not appear to reduce risk of death in HCM patients with nonsustained ventricular tachycardia (7). Furthermore, the survival of patients on monotherapy with high-dose adrenoceptor antagonist is just as good. Ventricular ectopics are rare before puberty (33), and particularly if they do not suppress at high heart rates they carry a risk of sudden death in patients with structurally abnormal hearts (34). Thus, it is probably appropriate to have a lower threshold for active treatment of ventricular ectopy in childhood HCM than in adults.
Current treatment practice
Since our analysis of the mortality results from the first two centers in 1997 (12), our threshold for active treatment is lower and, in addition to symptomatic patients and patients with dynamic outflow gradients or arrhythmias, all patients with a close family history of sudden death or with severe or rapidly progressing cardiac hypertrophy are recommended high-dose beta-blocker therapy. Propranolol remains the first line drug (6 to 23 mg/kg), except where there is a history of bronchospasm or hypoglycemia when we use metoprolol (6 to 12 mg/kg). We use atenolol (3.6 to 8 mg/kg) only in patients with sleep disturbance. High-dose beta-adrenoceptor blockade is associated with reduction in cardiac hypertrophy (12), which has been observed with propranolol, metoprolol and atenolol. Uniformly high levels of beta-blocking drugs throughout the 24 h are essential to maintain adequate heart rate control at all times, and where possible we use slow-release propranolol (the granules from an opened capsule can be fed to an infant, mixed in soft foods). We give plain propranolol three times daily and slow-release propranolol, atenolol and metoprolol twice daily.
Strong indications from this cohort study support the medical management of HCM presenting in childhood with HDβB therapy. This approach appears to confer a significant survival advantage, with no mortality in 26 patients treated for 199 patient years, and a 5 to 10-fold reduction in the hazard of death compared with the other treatments studied. These observations make a strong case for high-dose beta-blockers as the first-line treatment in symptomatic childhood HCM, and for a prospective randomized study of the use of high-dose beta-blocker therapy in the asymptomatic prepubertal child with HCM.
☆ This study was supported by a grant from the Lund–Oxford Biomedical Exchange Programme, Lund, Sweden, and by the Children’s Echocardiography Fund, Oxford Radcliffe Hospital Trust Fund, United Kingdom.
- congestive heart failure
- conventional-dose beta-adrenoceptor antagonist
- confidence interval
- hypertrophic cardiomyopathy
- high-dose beta-adrenoceptor antagonist
- left ventricle/left ventricular
- no specific therapy
- standard error
- Received April 8, 1998.
- Revision received June 15, 1999.
- Accepted August 12, 1999.
- American College of Cardiology
- McKenna W.J,
- Deanfield J.E
- Maron B.J,
- Henry N.L,
- Clark C.E,
- et al.
- Seiler C,
- Hess O.M,
- Schoenbeck M,
- et al.
- McKenna W.J,
- Oakley C.M,
- Krikler D.M,
- Goodwin J.F
- Frank M.J,
- Stefadouros M.A,
- Watkins L.O,
- et al.
- Östman-Smith I,
- Wettrell G
- Silverman N.H
- Feigenbaum H
- Maron B.J,
- Roberts W.C,
- Epstein S
- Brush J.E,
- Eisenhofer G,
- Garty M,
- et al.
- Shapiro L.M
- Alvares R.F,
- Goodwin J.F
- Matsubara H,
- Nakatani S,
- Nagata S,
- et al.