Author + information
- Received February 6, 2008
- Revision received May 23, 2008
- Accepted June 23, 2008
- Published online October 7, 2008.
- Michele Pasotti, MD⁎,
- Catherine Klersy, MD†,
- Andrea Pilotto, BS⁎,
- Nicola Marziliano, PhD⁎,
- Claudio Rapezzi, MD††,
- Alessandra Serio, MD⁎,
- Savina Mannarino, MD⁎,
- Fabiana Gambarin, MD⁎,
- Valentina Favalli, BME⁎,
- Maurizia Grasso, PhD⁎,
- Manuela Agozzino, MD⁎,
- Carlo Campana, MD‡,
- Antonello Gavazzi, MD∥,
- Oreste Febo, MD¶,
- Massimiliano Marini, MD#,
- Maurizio Landolina, MD‡,
- Andrea Mortara, MD⁎⁎,
- Giovanni Piccolo, MD‡‡,
- Mario Viganò, MD§,
- Luigi Tavazzi, MD‡ and
- Eloisa Arbustini, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Eloisa Arbustini, Centre for Inherited Cardiovascular Diseases, IRCCS Fondazione Policlinico San Matteo, Viale Forlanini 16, Pavia 27100, Italy
Objectives The aim of this study was to analyze the long-term follow-up of dilated cardiolaminopathies.
Background Lamin A/C (LMNA) gene mutations cause a variety of phenotypes. In the cardiology setting, patients diagnosed with idiopathic dilated cardiomyopathy (DCM) plus atrioventricular block (AVB) constitute the majority of reported cases.
Methods Longitudinal retrospective observational studies were conducted with 27 consecutive families in which LMNA gene defects were identified in the probands, all sharing the DCM phenotype.
Results Of the 164 family members, 94 had LMNA gene mutations. Sixty of 94 (64%) were phenotypically affected whereas 34 were only genotypically affected, including 5 with pre-clinical signs. Of the 60 patients, 40 had DCM with AVB, 12 had DCM with ventricular tachycardia/fibrillation, 6 had DCM with AVB and Emery-Dreifuss muscular dystrophy type 2 (EDMD2), and 2 had AVB plus EDMD2. During a median of 57 months (interquartile range 36 to 107 months), we observed 49 events in 43 DCM patients (6 had a later event, excluded from the analysis). The events were related to heart failure (15 heart transplants, 1 death from end-stage heart failure) and ventricular arrhythmias (15 sudden cardiac deaths and 12 appropriate implantable cardioverter-defibrillator interventions). By multivariable analysis, New York Heart Association functional class III to IV and highly dynamic competitive sports for ≥10 years were independent predictors of total events. By a bivariable Cox model, splice site mutations and competitive sport predicted sudden cardiac death.
Conclusions Dilated cardiomyopathies caused by LMNA gene defects are highly penetrant, adult onset, malignant diseases characterized by a high rate of heart failure and life-threatening arrhythmias, predicted by New York Heart Association functional class, competitive sport activity, and type of mutation.
Lamin A/C (LMNA) gene mutations have been causally linked to a variety of clinical phenotypes (1–3), the majority with cardiac involvement (1–5). Patients with LMNA gene mutations usually come to the attention of cardiologists because of dilated cardiomyopathy (DCM) plus atrioventricular block (AVB [MIM#115200]) (4–6), DCM plus AVB and Emery-Dreifuss muscular dystrophy type 2 (EDMD2 [MIM#181350]) (7–9), different types of myopathies (10,11), or overlapping syndromes including DCM, AVB, and/or arrhythmias (2,12–14). In the cardiology setting, patients diagnosed with DCM plus AVB constitute the majority of reported “cardiolaminopathies,” with AVB being a useful clinical marker to orient the search for LMNA gene mutations in patients with DCM (4,5,15). Sudden cardiac death (SCD) is a major clinical problem among all persons with LMNA mutations, 46% of whom have been reported to die suddenly (16,17). Information on incremental risk factors and prognostic stratification is limited both for persons with the mutations in general and for cardiolaminopathy patients in particular (11).
We investigated associations between genetic/nongenetic factors and long-term outcome in a cohort of DCM-affected families with LMNA mutations.
Setting and study design
This retrospective longitudinal study was conducted with a series of 27 consecutive families whose proband was diagnosed with a dilated cardiolaminopathy at our institution up to August 2006. The DCM was diagnosed based on World Health Organization criteria (18). Until 1994, we screened DCM patients and their family members when the pedigree suggested a familial disease. Since 1995, we have conducted serial family screening for all DCM patients (19). Criteria for search for the LMNA gene in DCM patients evolved during the study period according to current concepts (4–10,15–17,20,21). In brief, identification of the probands was based on the co-existence of isolated DCM with AVB (n = 21); DCM with AVB and EDMD2 (n = 4); and DCM and ventricular tachycardia/fibrillation (n = 2). We explored the associations between LMNA gene defects and phenotypic characteristics at presentation in our center and a history of highly dynamic competitive sports classified according to current guidelines (22). The patients were either phenotypically healthy or not aware of their status at the time of their athletic competitive activity. The mean interval that elapsed between the diagnosis and cessation of the competitive sport activities was 14.7 ± 10.2 years (median 12 years; range 0 to 36 years). Only 2 patients were agonistically active at the time of the diagnosis. We also assessed associations with outcome, and attempted prognostic stratification. The local ethics committee approved the study.
Family screening protocol
After giving informed consent, all probands and relatives received genetic counseling and clinical screening and underwent peripheral blood sampling for genetic testing. On confirming full comprehension of the screening protocol, each mutated person (or legal guardians of minors) received the written and signed report describing the identified mutation. Regular clinical monitoring was planned every 24 months for children and adolescents without instrumental abnormalities, and every 12 months for adults with electrocardiographic or echocardiographic abnormalities who did not fulfill criteria for the diagnosis of DCM or had DCM in functional class I and appropriate treatment, and at shorter intervals based on clinical needs. Clinical investigation included resting cardiologic evaluation, 12-lead electrocardiogram, 2-dimensional and echocardiography study (Doppler since 1978), 24-h Holter monitoring, effort test, and serum creatine phosphokinase determination. The left ventricular end-diastolic diameter was calculated according to the formula of Henry et al. (23). Neuromuscular assessment of probands and relatives was based on detailed clinical history, physical examination, serum creatine phosphokinase and serum lactate determination, and neuromyologic examination. When possible, we traced the clinical and pathology records and autopsy samples of deceased affected relatives.
The 12 exons of the LMNA gene were amplified from peripheral blood-derived genomic deoxyribonucleic acid by means of polymerase chain reaction, using primers derived from intronic sequences (5). The polymerase chain reaction fragments were analyzed by means of denaturing high-performance liquid chromatography using the Wave deoxyribonucleic acid fragment analysis system (Transgenomic, San Jose, California). Heteroduplex fragments were purified (QIAquick Kit, Qiagen, Santa Clarita, California) and then sequenced using a BigDye-terminator cycle sequencing system (ABI PRISM, Applied Biosystems, Foster City, California). The LMNA gene analysis was performed by direct bidirectional automated sequencing in case of probands with healthy parents, to exclude the rare possibility of recessive forms. The control series for each mutation is constituted of 196 normal persons. Reference sequences used in the study are as follows: LMNA gene: NCBI NC_000001; LMNA messenger ribonucleic acid: NCBI NM_170707; lamin A protein: NCBI NP_733821; LMNC messenger ribonucleic acid: NCBI NM 005572; lamin C protein: NCBI NP005563.
The 94 subjects with mutations were included in the statistical analysis. Mutations were categorized in two ways, based on type and position. Mutation types were classified as: missense; intervening sequence (or splice site) mutations predicting a skipped exon; or insertions/deletions and stop mutations (ins-del/stop). The mutations were also categorized by the two position clusters reported by Hegele (24) to be associated with different phenotypic classes: mutations upstream of the nuclear localization signal spanning residues 416 to 423 (where no mutations have been found to date) versus downstream mutations (Fig. 1). Descriptive statistics were computed as mean and standard deviation for continuous variables and as counts (percentages) for categorical variables. Median (interquartile range) follow-up duration was calculated according to the inverse Kaplan-Meier method. The Student t test, Fisher exact test, and test for trend were used to compare affected patients and phenotypically healthy mutation carriers at baseline. Binomial exact confidence intervals (CIs) were computed for the age-related penetrance.
Follow-up started at first clinical assessment at our centers and ended at the first occurrence of an event. Event-free survival analysis was performed within the overall study population of mutation carriers. Events were death from any cause, death from heart failure (HF), heart transplantation, and SCD, including appropriate implantable cardioverter-defibrillator (ICD) interventions. Rates per 100 person-years and their 95% binomial exact (95%) CIs were computed, as well as Kaplan-Meier cumulative survival (95% CI).
The log-rank test was used for univariable comparisons. Cox regression analysis was used to identify independent predictors of events and to attempt prognostic stratification. Hazard ratios and 95% CI were calculated. Variables reaching p < 0.2 at univariable analysis were included in the multivariable model, alongside mutation categories. Collinearity was observed between left ventricular (LV) ejection fraction and LV end-diastolic volume, and between age and LV ejection fraction and AVB. Therefore, LV ejection fraction was chosen based on clinical consensus as being the most common marker of LV function, and AVB was chosen as the most prevalent and specific marker of the disease. The Huber-White sandwich robust estimator of standard errors was used to account for intrafamilial aggregation of subjects. The proportional hazards assumption was verified by means of Schoenfeld residuals. The Harrell C statistic and the shrinkage coefficient were computed to assess model discrimination and calibration. Two-sided p values <0.05 were considered statistically significant. Stata 9.2 (StataCorp, College Station, Texas) was used for all computation.
Study population and baseline characteristics
The disease was proved to be familial autosomal dominant in 21 of the 27 families, likely familial autosomal dominant (based on pedigree and family history) in 5, and associated with a de novo mutation in 1. In 8 families, including that of the de novo case, the proband was the only living mutated member. In the other 19 families, the number of living mutated members ranged from 2 to 12. Of the 137 relatives, 67 (49%) carried the corresponding proband mutation. Thus, a total of 94 LMNA mutated subjects (i.e., 27 probands, 67 relatives) entered the main analysis. The 23 LMNA gene mutations identified in the 27 probands (6 novel and 17 known) are listed in Table 1.
Of the 60 mutated and affected patients, 58 had a phenotype including DCM and 2 (son and daughter of a patient with DCM, AVB, and EDMD2) had only AVB with EDMD2. Of the 34 mutated nonaffected subjects, 5 had conduction disturbances, supraventricular arrhythmias, or slightly increased LV diameter with normal LV ejection fraction but did not fulfill the diagnostic criteria for DCM; the remaining 29 were healthy. None had EDMD2. Table 2 summarizes the baseline characteristics of the entire study population. Among the 60 mutated affected patients, 12 had a pacemaker implanted before coming to our attention.
Fifty-nine mutations (63%) were missense, 14 (15%) were splice site, and 21 (22%) were ins-del/stop mutations. The 21 group of ins-del/stop mutations comprised 5 stop mutations (predicting a premature stop codon), 11 frameshift mutations, and 5 in-frame deletions predicting the loss of 2 residues. The grouping of in-frame deletions with out-of-frame and stop mutations was based on numerical reasons rather than on predicted molecular effects. Concerning position, 73 mutations (78%) were upstream of the nuclear localization signal and 21 (22%) were downstream (Table 1, Fig. 1). Estimated penetrance of the disease is summarized in Figure 2. Thirty mutated healthy subjects (88%) were under 40 years of age, as compared with 33 (55%) of the affected subjects (Fisher exact test p < 0.001). The overall distribution of missense, splicing, and ins-del/stop mutations did not differ significantly between nonaffected and affected persons (Fisher exact test p = 0.16), although mutations other than missense tended to be more frequent in affected than in nonaffected persons (44% vs. 23%, p = 0.073).
Outcome and prognostic stratification
The 94 mutated subjects were observed for a median follow-up time of 57 months (interquartile range 36 to 107 months). During follow-up, 20 patients received a pacemaker and 16 received an ICD implantation.
A total of 49 events occurred in 43 of the 60 affected patients (mean age at the event 42 ± 11 years; range 19 to 68 years), whereas no event was observed among the 34 mutated healthy subjects. Six nontransplanted patients had a later further event (heart transplantation, n = 3; death from congestive HF, n = 3) and were excluded from the statistical analysis. The mean follow-up was significantly shorter for younger healthy persons than for older mutated persons (32 ± 20 [range 1 to 84] months vs. 65 ± 78 [range 1 to 384] months, p = 0.017). The linearized event rate was 9.7 per 100 person-years (95% CI: 7.2 to 13.2). Cumulative event-free survival was 82% (95% CI: 71% to 88%) at 2 years and 60% (95% CI: 47% to 70%) at 5 years. The majority of the events were SCD/ventricular arrhythmias (27 of 43, 63%), comprising 15 SCD and 12 appropriate ICD interventions detected by the device (8 ventricular fibrillation and 4 sustained ventricular tachycardia). The remaining events were all related to end-stage HF: heart transplantation (n = 15) and death from congestive HF (n = 1). The highest percentages of total events and SCD-related events, including appropriate ICD interventions, were observed among subjects with intervening sequence mutations (64% and 57%), followed by ins/del-stop mutations (52% and 38%) and by the missense mutations (38% and 19%). The corresponding follow-up was 55 ± 98 (median 19) months, 52 ± 68 (median 33) months, and 49 ± 55 (median 36) months (p = NS).
At univariable analysis (Table 3), New York Heart Association (NYHA) functional class III to IV, any disease phenotype, atrioventricular conduction disturbances, LV ejection fraction <35%, LV end-diastolic volume >180 ml, and history of competitive sport were all associated with event occurrence.
In a multivariable Cox proportional hazards model, NYHA functional class III to IV and competitive physical activity both turned out to be strong independent predictors of events (about a 3- to 4-fold excess risk), whereas the type of mutation did not reach statistical significance (Table 4). To attempt prognostic stratification, we constructed a score based on the presence of these 2 risk factors. The cumulative event-free survival is illustrated in Figure 3. The hazard ratio for an event to occur was 5.20 (95% CI: 2.15 to 12.58, p < 0.001) when comparing patients with 1 versus 0 risk factors; 5.06 (95% CI: 2.45 to 10.43, p < 0.001) for 2 versus 1 risk factor; and 26.3 (95% CI: 8.32 to 83.11, p < 0.001) for 2 versus 0 risk factors.
The SCD/ventricular arrhythmia event rate among the affected patients was 6.27 per 100 person-years (95% CI: 4.29 to 9.14). On phenotypic grounds, 16 of the 27 patients who experienced SCD/ventricular arrhythmia had DCM plus AVB; 4 had DCM plus AVB plus EDMD2; 7 had DCM plus ventricular tachycardia/fibrillation. These 3 phenotypic groupings showed different SCD event rates (per 100 person-years): DCM plus AVB, 7.2 (95% CI: 4.4 to 11.8); DCM plus AVB plus EDMD2, 10.1 (95% CI: 3.8 to 26.9); and DCM plus ventricular tachycardia/fibrillation, 13.1 (95% CI: 6.2 to 27.4; p = 0.008). At univariable analysis (Table 5), DCM phenotypes, competitive sport, and mutation type were associated with SCD and appropriate ICD interventions. At Cox bivariable analysis, splice site mutations and competitive sport were predictors of SCD, 2.05 (95% CI: 1.01 to 4.17) and 3.8 (95% CI: 1.78 to 8.08), respectively (p < 0.001). For prognostic stratification purposes, we constructed a score based on these 2 risk factors. The cumulative event-free SCD survival is illustrated in Figure 4. The hazard ratio for an event to occur was 3.67 (95% CI: 1.53 to 8.77, p = 0.003) for 1 versus 0 risk factors; 2.09 (95% CI: 0.98 to 4.47, p = 0.056) for 2 versus 1 risk factor; and 7.68 (95% CI: 3.32 to 17.73, p < 0.001) for 2 versus 0 risk factors.
This large single-center cohort study reinforces the concept that DCM associated with LMNA mutations constitutes a highly penetrant, age-dependent, malignant disease characterized by high rates of major cardiac events. Survival analysis suggests the possibility of prognostic stratification based on phenotypic and genotypic characteristics.
From clinical suspicion to molecular diagnosis
To our knowledge, this is the largest reported cardiologic follow-up study of persons with LMNA gene mutations. Identification of the cohort (probands and family members) was facilitated by institutional protocols, which incorporated evolving concepts regarding serial family screening and peculiarities (“red flags”) that should arouse diagnostic suspicion of LMNA mutations. In line with current knowledge (1–6,10–13,15–17), identification of the probands was most often based on the combination of DCM and AVB in the context of non–X-linked and nonmatrilinear myopathy. Co-existent myopathy (EDMD2 or variable myopathy) (8-11) and increased serum creatine phosphokinase levels alone were also considered, after excluding an X-linked or matrilineal inheritance, especially in the presence of AVB, an unusual feature in X-linked DCM (25). In our cohort, most probands had familial disease, with only 1 case proven to be de novo. It should be noted that, due to selection bias considerations, including the characteristics of the institutional setting and the family screening protocol, the present study could not provide information regarding the prevalence of LMNA gene defects among patients with DCM.
Within our families, the overall penetrance was high and progressively increased with age: by 60 years, the penetrance was complete, and all mutated persons showed the phenotype, irrespective of mutation type. Among LMNA mutated subjects, DCM is rare under the age of 20 years. This observation is in line with the finding reported by Benedetti et al. (26) that onset of the cardiomyopathy phenotype tends to occur at a later age (adulthood) in comparison to the EDMD2 phenotype (early childhood). Of note, the distribution of nonmissense mutations in our cohort (44% in affected patients vs. 23% in unaffected subjects) is also broadly in line with the observation by Benedetti et al. (26) of a high prevalence in the cardiomyopathy phenotype of nonmissense mutations. The observation that the majority (78%) of our DCM-affected probands had mutations upstream of the nuclear localization signal spanning residues 416 to 423 is also in line with the study conducted by Hegele (24 using hierarchical cluster analysis, where these upstream mutations were strongly associated with class 1 (cardiac/myopathy phenotypes) with respect to class 2 laminopathies (partial lipodystrophy, progeria syndromes, and mandibuloacral dysplasia).
Our study is not informative regarding the reasons for the commonly observed phenomenon, described elsewhere (2,4,5,17), of early onset of AVB before DCM. We observed patients who first presented with AVB and later had DCM as well as patients who showed both DCM and AVB at the time of their first diagnosis (5). Prospective long-term follow-up of healthy mutated subjects should help elucidate the timing of expression of the phenotypic traits.
Outcome and prognostic stratification
Patients with LMNA gene mutations have high rates of major cardiac events (17). The major events recorded in our cohort were related to life-threatening arrhythmias or end-stage HF and always occurred in the context of DCM. The SCD-related events, including appropriate ICD interventions for life-threatening arrhythmias, were more frequent. The high rates of SCD and other SCD-related events are in line with the meta-analysis by van Berlo et al. (17), although we observed no major cardiac event before onset of DCM. However, the spectrum of our cardiac phenotypes reflects the DCM study setting, which does not include families in which myopathy was the isolated or major phenotype and which only partially overlaps with the setting of permanent cardiac pacing in patients without DCM. In a recent study (27) including 19 patients who underwent permanent pacing and ICD implantation solely on the basis of LMNA gene mutations associated with cardiac conduction defects and normal ventricular function, 8 of 19 (42%) received an appropriate ICD intervention, suggesting high risk of SCD before onset of HF. In our cohort, only 2 unrelated healthy young mutated subjects, both sons of patients with DCM plus AVB, had first-degree AVB with normal ventricular size and function. An important difference between our young healthy mutated subjects and the patients described by Meune et al. (27) is that our 2 young mutated persons with AVB did not yet require permanent pacing, and therefore, the indication for pacemaker and ICD implantation did not seem appropriate. Regarding myopathy traits, we observed overlapping phenotypes characterized by DCM plus AVB and EDMD2 in 4 of our families; however, the myopathy was mild, and the most relevant clinical problems were the DCM and arrhythmias. It is possible that, in the myology setting, patients with less cardiac remodeling are more common or that less attention is dedicated to early cardiac dysfunction, with SCD occurring in patients with apparently normal hearts.
At univariable analysis, several factors were associated with the risk of events both related to heart failure and to life-threatening ventricular arrhythmias. Surprisingly, however, at multivariable analysis, neither AVB nor pacemaker implantation turned out to be predictors of events, supporting the concept that AVB is a clinical marker of cardiolaminopathies, rather than an event predictor. In keeping with the results of the meta-analysis by van Berlo et al. (17), the pacemaker did not appear to protect from SCD. Of note, none of our patients with ICD died suddenly, but ICD allowed us to bridge 3 patients to heart transplantation.
Only 2 factors remained significant predictors at multivariable analysis: NYHA functional class III to IV (the strongest predictor of overall events) and a history of high dynamic competitive sports of at least 10 years. This interval is somehow arbitrary, as we did not have models or prior reference data for inherited DCMs. Obviously, the age influences the lower prevalence of athletes in the group of the younger healthy mutated persons, thus helping to explain the high association with the risk of events in the multivariable analysis. However, the perception of the negative effect of competitive sport on the disease evolution was a daily experience when screening and regularly monitoring families of our series. We are aware of the beneficial effects of sport on heart performance; however, the myocyte molecular and structural derangement associated with the LMNA gene defects could interfere with the potential benefit of physical training.
NYHA functional class III to IV was the strongest predictor of total events but not of SCD, which seemed to be more influenced by competitive sport and type of mutation. Competitive sport and presence of splicing mutations appeared to help the risk stratification of SCD at Cox bivariable analysis. The possible prognostic relevance of splicing mutations is an attractive hypothesis, and nonmissense mutations were associated with the highest rate of events. However, given the particular characteristics of the setting of this single study, caution is required before the presence of a splicing mutation is used for risk stratification in clinical practice. It is noteworthy that we were unable to detect any association between the mutation regions classified by Hegele (24) and outcome. It is likely that this classification is useful for predicting phenotypes, rather than their severity or related events.
Once DCM develops in persons with LMNA gene mutations, arrhythmogenic risk seems to increase, and ICD implantation should probably be performed earlier than suggested by current guidelines (28). We think that cardiolaminopathies with DCM phenotype deserve tailored clinical monitoring, including early indications for ICD implantation so as to prevent SCD and provide a possible bridge to heart transplantation in end stage HF. This concept fits well with proposals to bring forward ICD implantation in patients who are candidates for pacemaker implantation due to AVB (17,26,27). Given the increased risk of all major events and SCD in LMNA-positive subjects with clinically overt disease, highly competitive sports should be discouraged after detection of any phenotype. For mutated healthy persons, only further follow-up will give an answer on the suitability for agonistic activity. While awaiting these data, we suggest maintaining the sport activity at a nonagonistic level.
Long-term follow-up of a cohort of families with dilated cardiolaminopathies identified at a single center confirms their high risk of major events, particularly of SCD and ventricular arrhythmias. The emerging consensus in highlighting the risk of life-threatening arrhythmic events for persons with LMNA gene mutations is a warning that should lead to considering special indications for ICD implantation in this group of patients. We think that young healthy mutated patients should probably be advised to enjoy noncompetitive physical leisure activities or low dynamic, noncompetitive sport activities.
The authors are grateful to patients and families for their support of the activity of the Centre for Inherited Cardiovascular Diseases and for their helpful contribution to the critical discussion about their disease during the counselings.
Supported by grants “Ricerche Finalizzate e Correnti” from the National Ministry of Health to the IRCCS Policlinico San Matteo, and from Fondazione Cariplo, Milano, Italy.
- Abbreviations and Acronyms
- atrioventricular block
- dilated cardiomyopathy
- Emery-Dreifuss muscular dystrophy type 2
- heart failure
- implantable cardioverter-defibrillator
- insertions/deletions and stop mutations
- lamin A/C
- left ventricular
- sudden cardiac death
- Received February 6, 2008.
- Revision received May 23, 2008.
- Accepted June 23, 2008.
- American College of Cardiology Foundation
- Arbustini E.,
- Pilotto A.,
- Repetto A.,
- et al.
- Sanna T.,
- Dello Russo A.,
- Toniolo D.,
- et al.
- Brodsky G.L.,
- Muntoni F.,
- Miocic S.,
- Sinagra G.,
- Sewry C.,
- Mestroni L.
- Taylor M.R.,
- Fain P.R.,
- Sinagra G.,
- et al.,
- Familial Dilated Cardiomyopathy Registry Research Group
- Fernández X.,
- Dumont C.A.,
- Monserrat L.,
- Hermida-Prieto M.,
- Castro-Beiras A.
- Richardson P.,
- McKenna W.,
- Bristow M.,
- et al.
- Gavazzi A.,
- Repetto A.,
- Scelsi L.,
- et al.
- Mitchell J.H.,
- Haskell W.,
- Snell P.,
- Van Camp S.P.
- Henry W.L.,
- Gardin J.M.,
- Ware J.H.
- Arbustini E.,
- Diegoli M.,
- Morbini P.,
- et al.
- Zipes D.P.,
- Camm A.J.,
- Borggrefe M.,
- et al.