Author + information
- Received June 3, 1997
- Revision received April 3, 1998
- Accepted April 23, 1998
- Published online August 1, 1998.
- Elizabeth Goldmuntz, MD, FACCa,§,* (, )
- Bernard J Clark, MD, FACCa,§,
- Laura E Mitchell, PhD‡,§,
- Abbas F Jawad, PhD†,§,
- Bettina F Cuneo, MD, FACC¶,
- Lori Reed‡,
- Donna McDonald-McGinn‡,
- Peggy Chien‡,
- Jennifer Feuer‡,
- Elaine H Zackai, MD‡,§∥,
- Beverly S Emanuel, PhD‡,§ and
- Deborah A Driscoll, MD‡∥
- ↵*Address for correspondence to: Elizabeth Goldmuntz, MD, Division of Cardiology, Department of Pediatrics, The Children’s Hospital of Philadelphia, 34th Street & Civic Center Blvd., Philadelphia, PA, USA 19104
Objectives. This study was designed to determine the frequency of 22q11 deletions in a large, prospectively ascertained sample of patients with conotruncal defects and to evaluate the deletion frequency when additional cardiac findings are also considered.
Background. Chromosome 22q11 deletions are present in the majority of patients with DiGeorge, velocardiofacial and conotruncal anomaly face syndromes in which conotruncal defects are a cardinal feature. Previous studies suggest that a substantial number of patients with congenital heart disease have a 22q11 deletion.
Methods. Two hundred fifty-one patients with conotruncal defects were prospectively enrolled into the study and screened for the presence of a 22q11 deletion.
Results. Deletions were found in 50.0% with interrupted aortic arch (IAA), 34.5% of patients with truncus arteriosus (TA), and 15.9% with tetralogy of Fallot (TOF). Two of 6 patients with a posterior malalignment type ventricular septal defect (PMVSD) and only 1 of 20 patients with double outlet right ventricle were found to have a 22q11 deletion. None of the 45 patients with transposition of the great arteries had a deletion. The frequency of 22q11 deletions was higher in patients with anomalies of the pulmonary arteries, aortic arch or its major branches as compared to patients with a normal left aortic arch regardless of intracardiac anatomy.
Conclusions. A substantial proportion of patients with IAA, TA, TOF and PMVSD have a deletion of chromosome 22q11. Deletions are more common in patients with aortic arch or vessel anomalies. These results begin to define guidelines for deletion screening of patients with conotruncal defects.
Chromosome 22q11 deletions are present in the majority of patients with DiGeorge (DGS), velocardiofacial (VCFS) and conotruncal anomaly face (CTAFS) syndromes indicating that these conditions share a common genetic etiology (1–5). The major features of these syndromes include congenital heart disease, aplasia or hypoplasia of the thymus and/or parathyroid glands, palatal abnormalities, speech and learning disabilities and facial dysmorphia (6,7). The cardiac defects commonly seen in these disorders derive from the conotruncus, the embryonic aortic arches and the ventricular septum. In the original clinical description of these syndromes, interrupted aortic arch type B (IAA), truncus arteriosus (TA), and tetralogy of Fallot (TOF) comprised approximately 75% of the cardiac lesions found in DGS, whereas TOF, ventricular septal defects and right aortic arch were the most frequently reported defects in VCFS (8,9). All of the above lesions were reported in association with CTAFS (5).
Previous studies suggest that a substantial number of patients with conotruncal defects have a 22q11 deletion (10–15). However, these studies were limited by their relatively small sample size. The present study was designed to determine the frequency of the 22q11 deletion in a large, prospectively ascertained sample of patients with conotruncal defects on the cardiology service at a tertiary care center. Both primary diagnosis and associated cardiac anatomy were analyzed with respect to the deletion status. These results begin to provide cardiologists with guidelines for genetic testing and further evaluation of patients with conotruncal defects.
Patients with one of the following cardiac lesions were eligible for the study: IAA (type A, B or C), TA, TOF, transposition of the great arteries (TGA), double outlet right or left ventricle (DORV, DOLV respectively) and posterior malalignment type ventricular septal defect without interruption of the aortic arch (PMVSD). Subjects were recruited solely on the basis of cardiac findings. Extracardiac features were not considered in the recruitment process. Patients with IAA, TA and TOF were included in this study because, historically, these are the most common cardiac defects seen in DGS (8). Patients with PMVSD, TGA, DORV and DOLV were also studied because these lesions may be embryologically and genetically related to the defects seen in DGS. For the purpose of this study, DORV is defined only by the lack of fibrous continuity between the mitral and aortic valves in conjunction with an aorta arising more than 50% over the right ventricle, and therefore includes a wide spectrum of defects. In contrast, TOF is defined as the presence of an anterior malalignment type VSD with deficient subpulmonic conus and fibrous continuity between the mitral and aortic valves. PMVSD is defined as the posterior malalignment of the infundibular septum into the subaortic region and can be associated with aortic arch hypoplasia or coarctation but without complete interruption.
Patients with perimembranous VSD and isolated arch anomalies, which have been reported in association with VCFS, were not included as part of the present study. Patients with aneuploidy, such as trisomy 21, translocations or heterotaxy syndrome were excluded.
The majority of the study patients was identified prospectively from the elective cardiac catheterization or operative procedure schedules, or as newly diagnosed patients admitted to The Children’s Hospital of Philadelphia. The parents were approached prior to the procedure or while in the hospital, and were invited to participate in the research protocol. Recruitment at the time of a hospitalization or invasive procedure allowed for painless acquisition of a blood sample from indwelling central lines and increased the parental willingness to participate. Furthermore, the procedure schedules note the cardiac diagnoses but make no mention of extracardiac diagnoses, and thus help to minimize selection biases. A few older patients who were willing to undergo a venipuncture were recruited from the outpatient setting, but this patient pool was not systematically approached given the reluctance of most parents to allow their children to undergo venipuncture for research purposes. Given that all patients with one of the eligible cardiac diagnoses undergo surgery, the recruitment of hospitalized patients is unlikely to introduce significant bias in our sample. We attempted to recruit consecutive patients with these diagnoses, although the number of eligible patients often exceeded our resources. Thus, some patients were missed due to time and personnel constraints. At no time did the presence or absence of extracardiac features knowingly influence the decision to approach a family. A small number of patients (n = 26) were enrolled by one of our collaborators (BFC) following the same protocol. The study protocol was approved by the Institutional Review Board for the Protection of Human Subjects at The Children’s Hospital of Philadelphia. Written consent for each subject was obtained before enrollment into the study.
Two hundred fifty-one patients, including 39 patients previously reported, were prospectively enrolled in the study from November 1991 to June 1996 based on their cardiac diagnosis alone (10,15). One hundred thirty-five (53.8%) of the patients were male, 109 (43.4%) were female and 7 (2.8%) were unknown. There were 168 (66.9%) whites, 45 (17.9%) African-Americans, 9 (3.6%) Hispanics, 4 (1.6%) Asians, 2 (0.8%) Native-American Indians, and 23 (9.2%) of other or unknown race.
Cardiac evaluation and diagnosis
Cardiac anatomy was diagnosed by echocardiography, cardiac catheterization or cardiac magnetic resonance imaging studies. When available, the surgical operative note was reviewed (213/251 cases overall and 6/6 cases with PMVSD). The data were reviewed by two cardiologists (E.G., B.J.C.) and entered into an Ingres database. Only those patients with complete anatomic description of their cardiac defects were included in this study. The primary diagnoses included IAA, TA, TOF, PMVSD, DORV, DOLV or TGA. All associated cardiac features were entered as secondary diagnoses including cardiac segments, side of the aortic arch, aortic arch anomalies, pulmonary artery anomalies, presence of aortopulmonary collaterals and systemic or pulmonic venous anomalies.
Deletion status was determined in the initial 17 patients by quantitative Southern blot hybridization as previously reported (10). The remaining patients were evaluated for the presence of a deletion by fluorescence in situhybridization (FISH). Metaphase chromosomes from either peripheral blood lymphocytes, lymphoblastoid cell lines, cultured amniocytes or chorionic villi were cohybridized with the cosmid probe N25(D22S75) from within the DiGeorge chromosomal region (DGCR) and a control probe [cos 82 or pH17(D22S39) as supplied by Oncor] mapping to the distal long arm of chromosome 22 (Fig. 1). A cosmid probe for locus D22S66 was used to evaluate patients with TGA for smaller deletions within the DGCR.
The frequency of the chromosome 22q11 deletion was calculated for the total sample of patients, as well as for the selected subgroups of patients. Differences in the frequency of this deletion, between subgroups of patients defined by either primary or secondary cardiac diagnoses, were compared by chi-square analysis, Fisher’s exact test or the odds ratio.
Multiple logistic regression analyses were used to determine whether deletion status (that is, deleted versus non-deleted) is independently related to both the primary and secondary cardiac diagnoses. These analyses were restricted to data from individuals with one of four primary diagnoses (IAA, TA, TOF and PMVSD) and the secondary cardiac diagnoses were grouped into four, non-overlapping categories: normal left aortic arch (i.e., no secondary cardiac anomaly), right aortic arch with mirror image branching, left aortic arch with at least one vessel abnormality and right aortic arch with at least one vessel abnormality.
Three logistic regression models were fit to the data. The full model (model 1) included three dummy variables that defined the four primary diagnoses, and three dummy variables that defined the four secondary diagnoses. Two reduced models, one including only the dummy variables for the primary diagnoses (model 2), and the other including only the dummy variables for the secondary diagnoses (model 3) were also fit to the data. The fit of each reduced model, relative to that of the full model, was evaluated by use of the likelihood ratio test (that is, twice the difference between the log likelihood of the full and reduced model). This test is distributed approximately as a chi-square test, with degrees of freedom determined by the difference in the number of parameters estimated by the two models.
Comparison of the full model (model 1) with model 2 tests whether secondary cardiac diagnoses are significantly related to deletion status, independent of the effects of the primary diagnoses. In contrast, comparison of the full model with model 3 tests whether primary cardiac diagnoses are significantly related to deletion status, independent of the effects of the secondary diagnosis.
Statistical analyses were conducted using SAS version 6.10 and Epi Info version 6.03 (16,17).
Deletion frequency by primary cardiac diagnosis
A total of 251 patients with one of seven conotruncal defects were enrolled in the study and evaluated for a 22q11 deletion at locus D22S75 (N25) (Fig. 1). Forty-five patients (17.9%, 95% C.I. 13.2 to 22.7) were found to have a deletion. The proportion of patients with a 22q11 deletion varied with the primary cardiac diagnosis (Table 1). Deletions were most common in patients with one of the three cardiac defects seen most commonly in DGS including 50.0% (12/24) of patients with IAA, 34.5% (10/29) of patients with TA, and 15.9% (20/126) of patients with TOF, as well as 33.3% (2/6) of patients with PMVSD. Deletions were only found in patients with IAA type B, although only three patients with IAA type A were evaluated. There was no statistical difference in the deletion frequency between patients with TOF/pulmonary atresia, TOF/pulmonary stenosis or TOF/absent pulmonary valve (Fisher’s exact test, p = 0.55). Only one of 20 patients with DORV had a 22q11 deletion.
Deletions at locus D22S75 (N25) were not detected in 45 patients with either d- or l-TGA, with or without a ventriculoseptal defect (Table 1). To assess whether some of these patients might have a smaller deletion distal to D22S75, 21 of the 45 patients were studied by FISH using a probe for locus D22S66, which maps approximately 550 kb distal to D22S75 (Fig. 1). None of the 21 TGA patients studied was deleted at locus D22S66.
Deletion frequency and associated vascular anomalies
The relationship between 22q11 deletions and associated vascular anomalies was evaluated in the subset of patients with the highest deletion frequencies (IAA, TA, TOF and PMVSD). For the purpose of these analyses, associated vascular anomalies included right aortic arch, cervical aortic arch, aberrant right or left subclavian artery, aorticopulmonary collaterals, or absent or discontinuous branch pulmonary arteries. Three comparisons were designed to assess the relationship between deletion status and associated vascular anomalies. Patients without any associated vascular anomaly (that is, patients with a normal left aortic arch) were compared to those with any type of arch or vessel anomaly (Table 2, Comparison I). Patients with at least one vascular anomaly were significantly more likely to have a deletion of chromosome 22q11 than were patients without any vascular anomaly (odds ratio = 4.4, 95% CI 2.0 to 9.9).
To determine whether deletion status is associated with arch sidedness in the absence of other vessel anomalies, patients with a right aortic arch and mirror image branching were compared to patients with a left aortic arch and normal branching pattern (Table 2, Comparison II). Deletions were significantly more common in patients with a right aortic arch than a left aortic arch (OR = 3.1, 95% CI 1.2 to 8.3).
The relationship between deletion status and vessel anomalies, ignoring arch sidedness, was also assessed (Table 2, Comparison III). Patients with either a normal left aortic arch or right aortic arch and mirror image branching pattern were compared to those with either a left or right aortic arch and additional vessel anomalies. Deletions were significantly more common in patients with an additional vessel anomaly than in those without such abnormalities (odds ratio = 4.2, 95% CI 1.8 to 9.6).
These analyses were repeated in the subgroup of patients with TOF, the largest subgroup sharing the same primary diagnosis. The results were not markedly different from those described above (Table 2). However, none of the comparisons was statistically significant in this subgroup of patients (that is, the 95% confidence interval for the odds ratio included unity). The relatively small number of patients with IAA, TA and PMVSD precluded meaningful analyses of these subgroups.
Multiple logistic regression analyses
Multiple logistic regression analyses were undertaken in an attempt to determine whether the association between deletion status and associated vascular anomalies was independent of the association between deletion status and the primary cardiac diagnosis. These analyses were restricted to patients with a primary diagnosis of IAA, TA, TOF or PMVSD, and patients were classified into four, nonoverlapping groups on the basis of arch sidedness and the presence/absence of at least one vessel abnormality: normal left aortic arch, right aortic arch with mirror image branching, left aortic arch with at least one vessel abnormality and right aortic arch with at least one vessel abnormality.
The logistic regression model that included both the primary and secondary diagnoses as predictor variables provided a significantly better fit to these data than did either the model which included only the primary cardiac diagnosis (chi-square test with three dummy variables = 22.0, p = 0.00006), or the model which included only the secondary diagnosis (chi-square test with three dummy variables = 17.0, p = 0.0007) (Appendix). These results indicate that both the primary and secondary cardiac diagnoses are independently related to deletion status.
The logistic regression equation for the model, which included both primary and secondary cardiac diagnoses, was used to predict the proportion of patients with specific combinations of primary and secondary diagnoses who are expected to have a deletion of chromosome 22q11 (Table 3). It can be seen from Table 3that, within each primary diagnosis category, the expected proportion of patients with a deletion of chromosome 22q11 increased in the presence of the secondary diagnosis.
Deletion frequency and primary cardiac diagnosis
This analysis was designed to determine the frequency of 22q11 deletions in cardiac patients based on their primary cardiac diagnoses and associated vascular features. These findings may assist the clinician in deciding which patients warrant further genetic and clinical evaluation for the 22q11 deletion syndrome. This study analyzes the largest sample of patients ascertained solely on the basis of cardiac status reported to date. At the present time, the deletion frequencies reported in this study serve as the best estimate of the frequency in patients with the specific cardiac diagnoses. However, this report is not a population-based study and the reported deletion frequency may not represent the absolute prevalence of the deletion in the patient population with conotruncal congenital heart disease.
Deletions of 22q11 were detected in 17.9% of the 251 patients ascertained with one of seven conotruncal defects: IAA, TA, TOF, PMVSD, TGA, DORV and DOLV. Thus, this study confirms that 22q11 deletions are important in the etiology of conotruncal defects. The deletion frequency varied with the primary diagnosis and was highest in patients with IAA and TA (50.0% and 34.5%, respectively). These findings are consistent with the clinical report by Van Mierop and Kutsche (8). In our sample, 15.9% of patients with TOF have a deletion, which is in agreement with other studies reporting that 8 to 23% of TOF patients have a 22q11 deletion (11–13). Our results indicate that deletions of 22q11 are as common in patients with TOF/pulmonary atresia as those with TOF/pulmonary stenosis or TOF/absent pulmonary valve, in contrast to another study suggesting that the deletion frequency is higher in patients with TOF/absent pulmonary valve (18). However, our sample size of 3 patients with TOF/absent pulmonary valve is small.
Two of six patients with PMVSD had a deletion, a finding that has not been previously reported and extends the spectrum of cardiac pathology associated with a 22q11 deletion. In conjunction with previous reports, this finding suggests that multiple abnormalities of the infundibular septum may occur in association with the 22q11 deletion syndrome. The reported abnormalities of the infundibular septum include anterior malalignment (as seen in TOF), posterior malalignment (as seen in isolation or in conjunction with IAA) and infundibular hypoplasia (as seen in TOF) (19,20).
Only one of the 20 patients with DORV in this study had a 22q11 deletion. Takahashi and colleagues (12)reported that none of eight patients with DORV had a deletion. These reports together suggest that 22q11 deletions are relatively infrequent in this subgroup. The only patient in the present study with DORV and a deletion had a subaortic ventricular septal defect, right aortic arch and isolated left pulmonary artery. This lesion is reminiscent of TOF with multiple arch anomalies and may explain why this subject was the only DORV patient with a deletion. Given the anatomic variability within the category of DORV, a small subgroup of these patients may have a 22q11 deletion. Hence, further investigation of a larger sample of patients with DORV with detailed anatomic description is warranted.
In the present study, none of the 45 patients with either d- or l-TGA were found to have a 22q11 deletion using the commercially available probe N25 (D22S75). Our results are similar to those reported in a smaller study (n = 16) by Takahashi and colleagues (12). In contrast, three patients with TGA and a ventricular septal defect have been reported in association with unbalanced translocations resulting in deletion of this region (21). In addition, Melchionda and colleagues (22)demonstrated a deletion at locus D22S134 (KI-182/HP500) within the commonly deleted region in 12.5% (4 of 32) of patients with TGA. The differences in these studies may reflect the small sample size or unappreciated differences in the patient populations. Based on the published data, it appears that the proportion of TGA patients with a 22q11 deletion is likely to be relatively low.
Deletion frequency and associated vascular abnormalities
Our data indicate that the deletion frequency is higher in patients with the secondary diagnosis of any arch or vessel anomaly as compared to those with a normal left aortic arch. We further investigated whether arch sidedness or the presence of other vessel anomalies was associated with deletion status. Patients with a right aortic arch were more likely to have a chromosome 22q11 deletion than patients with a normal left aortic arch. The frequency of deletions was also significantly higher in patients with a vessel anomaly regardless of arch sidedness.
Multiple logistic regression analyses indicated that the association between secondary cardiac abnormalities (arch and/or vessel abnormalities) and deletions of chromosome 22q11 are independent of the association between this deletion and the primary cardiac abnormality. Based on the full logistic regression model, the probability of having a deletion varied with the primary diagnosis, and within each primary diagnosis category, the probability of having a deletion varied with the secondary diagnosis. Since the logistic regression model is based on a relatively small sample of patients, the predicted probabilities of having a deletion of chromosome 22q11 are likely to be imprecise and should not be used for genetic counseling or patient management. However, the observed association between arch/vessel anomalies and chromosome 22q11 deletions provide insight into the pathogenesis of the cardiac anomalies seen in the 22q11 deletion syndrome.
Our findings are similar to those reported by Momma and colleagues (19,23)who compared the additional cardiac features in patients with TOF (PS or PA) with a deletion to those without a deletion. In both studies, patients with a deletion were found to have an anomaly of the pulmonary arteries, aortic arch or its major branches more frequently than patients without the deletion.
Deletion screening and noncardiac features
This investigation evaluates the likelihood of identifying the patient with a 22q11 deletion based solely on cardiac anatomy. Patients with 22q11 deletions may present with other important findings such as palatal abnormalities, craniofacial dysmorphia, absent thymus, hypocalcemia, speech and learning difficulties or neuropsychological disorders (24). Patients with conotruncal cardiac disease should be examined for the presence of noncardiac abnormalities which may assist in identifying those patients who are likely to have a deletion. However, subtle noncardiac features may not be apparent in infancy when the vast majority of patients are diagnosed with congenital heart disease. For example, several investigators have reported classifying a patient as nonsyndromic only to be surprised at the finding of a deletion (11,13,14). In these studies, reexamination of the patient at an older age revealed features consistent with VCFS that were not appreciated on earlier examinations. In contrast, patients may have features suggestive of VCFS without a 22q11 deletion. For example, in our original report, 5 of 12 patients without a 22q11 deletion had additional findings. Likewise, in the report by Webber et al. (14), many patients were classified as having features of VCFS but were not found to have the deletion. Some of the classic features are age-dependent while others, such as facial dysmorphia, are more subjective. Therefore, the present study design did not consider findings other than cardiac anatomy in the analysis, and demonstrates that cardiovascular anatomy alone can be used to predict which patients are more likely to have a deletion. Additional studies are in progress to assess how inclusion of objective diagnoses such as thymic aplasia, palatal abnormalities or symptomatic hypocalcemia in the neonate alters the probability of finding a deletion. Only a longitudinal study that follows the infant through school age and perhaps into adulthood will identify the presence of all potential noncardiac features (such as learning disabilities, hypernasal speech, neuropsychiatric disorders or facial dysmorphia) and determine whether any patients with truly isolated conotruncal defects have a 22q11 deletion. In the future, these analyses may enable us to better identify which cardiac patients should be tested for the deletion.
In summary, this study demonstrates that cardiac patients with IAA, TA and TOF frequently have a 22q11 deletion. Despite the paucity of predictive outcome information, we believe that it is important to recognize the patient with a 22q11 deletion at an early age to allow for the diagnosis and treatment of associated medical problems such as feeding disorders, hypocalcemia, immune deficiencies, neurodevelopmental difficulties, palatal abnormalities and speech problems. In addition, knowledge of the deletion status will influence the genetic counseling provided to the family and other family members. Thus, at this time, we recommend that all newly diagnosed infants with IAA, TA and TOF be evaluated for the presence of a 22q11 deletion. Deletion testing of school age children with these cardiac defects is recommended in the presence of additional features of VCFS. Patients with a PMVSD should also be considered for deletion screening. Our data suggest that deletions in patients with TGA and DORV are infrequent in contrast to other reports (21,22). Therefore, until additional studies are completed, we recommend testing patients with TGA or DORV for a deletion when other features of VCFS are diagnosed. The presence of extracardiac features commonly seen in the 22q11 deletion syndrome should prompt a genetic evaluation and molecular-cytogenetic studies in any cardiac patient. Finally, studies integrating noncardiac features into the regression analysis of deletion probability are underway and may help to develop appropriate guidelines for deletion screening. Additional studies looking at the cost effectiveness and long-term benefit of deletion testing may also be helpful in the development of guidelines for testing, as will the long-term follow-up and evaluation of patients with conotruncal malformations.
The authors wish to thank the members of the Divisions of Cardiology and Cardiothoracic Surgery as well as the Staff of the Cardiac Intensive Care Unit at The Children’s Hospital of Philadelphia for assistance with patient recruitment, Yvonne Tatsumura for technical assistance, Colleen Oates for secretarial assistance; and ONCOR, Gaithersburg, MD for providing the N25 and pH17 probes.
Logistic regression analyses were used to assess the association between deletion status and both primary and secondary cardiac diagnoses. Three models were fit to the data. The full analysis model included 6 variables: three dummy variables which defined the four primary diagnoses (i.e., IAA, TA, VSD and TOF), and 3 dummy variables which defined the four secondary diagnoses (i.e., none, right aortic arch with mirror image branching, left aortic arch with at least one vessel abnormality, and right aortic arch with at least one vessel abnormality). Two reduced models, one including only the three dummy variables defining the primary diagnosis, and one including only the three dummy variables defining the secondary diagnosis, were also fit to the data. The fit of the reduced models to the data, relative to the fit of the full model was evaluated by use of the likelihood ratio test—that is, twice the difference between the log likelihood of the full and reduced models. This test is distributed approximately as a χ2, with the degrees of freedom determined by the difference in the number of parameters estimated by the two models.
The model which included both the primary and secondary diagnoses provided a significantly better fit to these data than either of the reduced models (Table 4), indicating that both primary and secondary cardiac features are related to a patients deletion status. The odds ratios and associated 95% confidence intervals obtained under the full model are provided in Table 5.
☆ This study was supported by a SCOR grant from the Heart, Lung and Blood Institute (HL51533) and HL03191 (EG).
- conotruncal anomaly face syndrome
- DiGeorge chromosomal region
- DiGeorge syndrome
- double outlet left ventricle
- double outlet right ventricle
- fluorescence in situhybridization
- interrupted aortic arch
- posterior malalignment type ventricular septal defect
- transposition of the great arteries
- truncus arteriosus
- tetralogy of Fallot
- velocardiofacial syndrome
- Received June 3, 1997.
- Revision received April 3, 1998.
- Accepted April 23, 1998.
- American College of Cardiology
- Driscoll D.A.,
- Salvin J.,
- Sellinger B.,
- et al.
- Burn J.,
- Takao A.,
- Wilson D.,
- et al.
- Shprintzen R.J.,
- Goldberg R.B.,
- Young D.,
- Wolford L.
- Goldmuntz E.,
- Driscoll D.,
- Budarf M.L.,
- et al.
- Trainer A.H.,
- Morrison N.,
- Dunlop A.,
- Wilson N.,
- Tolmie J.
- Cuneo B.F.,
- Langman C.B.,
- Ilbawi M.N.,
- Ramakrishnan V.,
- Cutilletta A.,
- Driscoll D.A.
- SAS Institute Inc
- Dean A.G.,
- Dean J.A.,
- Coulombier D.,
- et al.
- Momma K.,
- Kondo C.,
- Matsuoka R.
- Driscoll D.A.,
- Emanuel B.S.