Main

Since the first description of DBA as a congenital erythroblastopenia in the 1930s (1, 2), numerous cases of this disease have been reported; a comprehensive review covering >200 cases was published in 1988 (3). More recent reports have analyzed clinical data from either a single clinical center or national registries and have provided insights into clinical presentation, biologic features, and outcome of DBA (4–6). In the present study, we present the analysis of clinical data from a series of 229 DBA patients from France, Germany, and other European countries. All of these individuals had regular follow-up, and extensive biologic and clinical information was available for analysis of the progression of the disease and clinical outcome. In the present article, we highlight clinical presentation as well as long-term prognosis and response to the different treatment schedules. A vast majority of the severe complications encountered are of iatrogenic origin, related either to transfusion dependence or to long-term steroid therapy. Alternative treatments including BMT seem appropriate in this group of severely affected individuals.

Epidemiologic evaluation of DBA required a systematic and intensive search to identify all DBA patients born in France since 1983 and resulted in the identification of 72 patients who represent all or almost all French individuals with DBA born since this period. On the basis of this finding, we estimate the incidence of the disease to be 7.3 cases per million live births over the 13-y study period. No significant gender difference was noted in either presentation features or clinical outcome. Prevalence of malformations observed in our study is similar to that reported in previous studies. Importantly, our analysis suggests significant relationships between clinical manifestations, biologic features, familial history, and disease evolution.

METHODS

Registry.

The French DBA registry was initiated in 1984 by the DBA working group of the Société d'Hématologie et d'Immunologie Pédiatrique (SHIP). Since 1995, exhaustive coverage of pediatric French cases has been achieved through distribution of inclusion files to all French hematologists and pediatric departments. Annual notices were sent to collect new cases and follow-up data from previously registered patients. The French registry has been approved by the Commission Nationale Informatique et Libertés. Patients were recruited from 80 pediatric and/or hematology departments throughout France and its overseas territories. One hundred fifty-nine DBA cases were referred to the registry by medical groups from France, whereas eight others were referred by physicians from Belgium (seven patients) and Switzerland (one patient). Seventy German patients were added to the registry in 1998 through the DBA working group of the European Society for Pediatric Hematology and Immunology (ESPHI). Of the 237 patients initially referred for inclusion in the DBA registry, eight were excluded due to ambiguous or erroneous diagnoses (congenital dyserythropoiesis, two; transient erythroblastopenia of childhood, two; aplastic anemia, one; metabolic disease, one; dysmyelopoiesis, two). Thus, 229 DBA patients were finally available for detailed analysis.

A standardized questionnaire was used to collect data on each patient. It included information on birth history, physical examination, hematologic and biochemical data, response to therapeutic procedures, and a standardized clinical assessment of dysmorphic features based on a previously published range of malformations (3). Follow-up information was ascertained on the basis of annual patient clinical checkup and clinical records detailing growth parameters, CBC, all therapeutic procedures since the last evaluation, transfusion program, and treatment complications. Final analysis of the assembled data was performed in September, 1998. Descriptive statistics on clinical presentation, biologic features, evolution data, and prognosis factors were performed on the whole registry population, whereas epidemiologic analysis was restricted to the subgroup of patients born in France.

Diagnosis and presentation.

Diagnosis criteria included aregenerative anemia related to pure erythroblastopenia (≤5% of nucleated cells in the bone marrow are erythroblasts, without dysmorphic features) in an otherwise normal bone marrow. Chronic parvovirus B19 infection was ruled out by serology and/or PCR on blood and/or bone marrow and Fanconi's anemia by chromosome breakage study (7). These analyses were systematically performed before inclusion in the registry for all patients born since 1985; this information is being progressively updated when not available for older patients. Both diagnoses were thus ruled out in 47% of the registry population screened at the time of this analysis. All physical abnormalities were included in the database, enabling us to calculate the number of malformations per organ and the number of malformations for each individual (Table 1). Physical characteristics, when related to either premature birth or maturation delay or when frequently found in otherwise healthy individuals (i.e. foramen ovale persistence, hip dysplasia, testis migration defect), although recorded in the database, were excluded from the determination of the clinical phenotype and from calculations on prevalence of malformations associated with DBA. Weight and height at birth were compared with intrauterine growth charts according to gestational age (8, 9). Hypotrophy was defined as birth weight below the 10th percentile of body weight for gestational age and statural retardation as birth length below the 10th percentile of body length for gestational age.

Table 1 Malformations in DBA at presentation (n= 229) Note that total number of patients exhibiting malformations in one organ or category is less than the sum of all malformations encountered in the same category, as some patients may present more than one dysmorphic feature.
Full size table

Growth parameters at diagnosis or during evolution were transformed into SD from the mean [SD score or z score (10)] of the considered variable (body weight, body length, body mass index) according to mean and SD values for age and sex by using reference tables (11, 12). Association between malformations and descriptive variables at diagnosis was searched for by univariate analysis, and all parameters with a level of significance below 0.05 were included in a logistic regression model.

Disease evolution.

Hematologic outcome was assessed after completion of the first steroid course and at last follow-up. An intermediate evaluation of outcome was also used to describe variations in therapeutic response. Patients were classified into the following two groups:1) SR under long-term steroid therapy, making it possible to spare transfusions, and 2) TD patients who were steroid nonresponsive, had experienced a partial response to steroid therapy as previously described (7), or in whom steroid therapy had to be discontinued because of unacceptable side effects. Patients were considered free of treatment when persistently independent from both transfusion and steroid therapy for at least 6 mo. Because BMT dramatically transforms the course of the disease, the 13 patients who received allogenic bone marrow were not considered in the analysis of prognosis factors but consequently were analyzed separately.

Prognosis factors.

As long-term transfusion dependence has been shown to carry an unfavorable prognosis compared with long-term steroid sensitivity, variables in the outcome analysis included 1) outcome after first steroid course (responders versus nonresponders) and 2) a two-classes variable, “long-term prognosis.” Patients either free of treatment or persistently transfusion independent on steroid therapy at the last follow-up were considered the good prognosis group. The poor prognosis group included patients on a regular transfusion program at last follow-up or those who died after several years of transfusion dependence. Univariate association between these two outcome variables and potential explicative parameters included hypotrophy at birth, premature birth, diagnosis evoked before/after age 2, age at presentation, familial history of DBA, presence/absence of any malformation, presence/absence of malformations in each of the following categories: head, eye, neck, thumb, heart, bone, urogenital tract, miscellaneous. Other parameters at diagnosis included hypotrophy and statural delay (z score for, respectively, body weight and body length below/above −2 SD) and thrombocytopenia or thrombocytosis (platelet count below 150 × 109/L or above 450 × 109/L, respectively).

Epidemiology.

Incidence of DBA was calculated by comparing the number of DBA patients born in France between 1983 and 1995 with the total number of live births in France during the same period: 9 777 456 (INSEE, Paris, France). To analyze a possible effect of seasonality on DBA (5), the month of conception was calculated from birth date and term of pregnancy when available (n= 175). To search for a seasonality effect on potentially affected fetuses or maternal miscarriage, the distribution of month of birth of DBA patients was compared with monthly variations in the total number of births in France (INSEE, Paris, France). The observed number of cases for each month of the year was compared with the expected number of cases according to incidence of the disease and monthly rate of birth. Comparison was performed through χ2 analysis. Prevalence of malformations was calculated according to the month of birth and the month of conception. The same analysis was also performed on a 3-mo grouping basis according to all combinations of contiguous months.

Statistical analysis.

Descriptive statistics are presented as percentages and mean ± 1 SD. Crude and specific prevalence rates were calculated as the number of patients with the corresponding feature divided by the total number of patients in the relevant subgroup. The significance of observed differences was tested using the χ2 statistic and, when appropriate for small samples, the Fisher exact test on a bilateral formulation. Kruskal-Wallis H nonparametric test was used for comparison of age at diagnosis, as variance within groups seemed to be different. All p values were 2-sided, and a p value of 0.05 for univariate analysis was considered significant according to the large number of tests performed. A p value of 0.01 was chosen as significant for seasonality analysis, as multiple tests and grouping of data were used. To determine predictive factors for favorable outcome in the whole group or in the subgroup of steroid-resistant patients, multivariate logistic regression models were developed. The models included any variable with univariate relationship with a statistical test below 0.05 to this outcome variable. The results were reported as odds ratios (OR) with 95% confidence interval. Actuarial survival was calculated only for patients born after 1977, because patients included in the registry in adulthood can lead to a recruitment bias and cannot be considered representative of the whole adult DBA population. Survival was analyzed through the Kaplan-Meier technique on the whole group and on two separate groups according to the response to the first steroid course. The difference between steroid-responsiveness groups was assessed by log-rank test. Statistical analysis was performed using Epi-Info 6.04b (13) or Statistica 4.0 B (Statsoft, Inc.) software.

RESULTS

Diagnosis and presentation.

Among the 229 individuals in the registry, diagnosis of DBA could be established at birth in 30 cases (13.1%), before 1 y of age in 202 cases (88.2%), before 2 y of age in 217 patients (94.8%), and before 5 y of age in 226 patients (98.7%). Median age at presentation was 3 mo (Fig. 1). As diagnosis of DBA after the age of 2 is questionable (7), the 12 patients whose anemia was revealed late were specifically studied; one is a familial case from a dominant family in which DBA manifested in other individuals of the family before the age of 2, three are TD, and three others are long-term steroid dependent. After being weaned from both steroids and transfusions, the five remaining patients exhibit a persistent macrocytosis, with one individual also exhibiting a persistent increase in erythrocyte adenosine deaminase activity (eADA). Parvovirus B19 serology and PCR studies did not reveal a viral infection in any of these patients. Thus, biologic features and clinical outcome data are consistent with the diagnosis of DBA in these 12 cases with a delayed diagnosis.

Figure 1
figure 1

Age at presentation of DBA. First signs of anemia were discovered earlier in patients with a severe outcome.

Full size image

The male:female ratio was not significantly different from 1 (1:1.1). Premature birth (before 37 wk of gestational age) occurred in 19.9% of cases, hypotrophy at birth was observed in 28.4% of the patients, and statural retardation in 22.1% of the patients. Birth weight was <2500 g in 26.5% of the patients. At least one malformation was observed in 93 patients (40.6%), whereas 54 patients were affected with two or more malformations (23.6%). Details on all malformations and prevalence for each group of malformations are shown in Table 1.

CBC was not available for all DBA patients at diagnosis or before the first transfusion. Macrocytosis based on mean corpuscular volume reference values for the appropriate age range was observed at presentation in 22% of the patients for whom the initial data were available but was present in 59.6% of 92 non-TD patients at last follow-up, a feature reported previously (4). Erythrocyte adenosine deaminase activity was found to be elevated in 28 of 34 non-TD patients, as previously described (14). At diagnosis, moderate neutropenia below 1.5 × 109/L was observed in 24.4% of cases, thrombocytopenia with platelet count below 150 × 109/L in 9.4%, and thrombocytosis with platelet count above 450 × 109/L in 41.2%. By univariate analysis, the presence of at least one malformation was significantly associated with two variables: short stature and absence of hypotrophy at diagnosis. We forced the existence of a familial history as a third variable in the regression model. A logistic regression model including these variables showed all three to remain significantly and independently associated with the presence of malformations (n= 118, r= 0.56). Presence of malformations is significantly and independently associated with familial history of DBA, short stature at diagnosis, and absence of hypotrophy at birth. Final stature was observed to be below −2 SD in 51.9% of the patients who reached adulthood.

Familial history.

Of the 229 patients in the registry, 195 were sporadic cases, whereas dominant mode of inheritance was observed for 34 patients from 16 families. Consanguinity was noticed in three pedigrees. Analysis of pedigrees of dominant families did not reveal any modification in sex ratio in affected or obligate carrier parents (male, six; female, five) compared with DBA cases (male, nine; female, 15). In dominant families, comparison of outcome between 23 DBA children and their 11 corresponding affected parents did not show any significant difference between generations. At last follow-up, seven children and seven affected parents were free of treatment; 11 children and four parents were steroid dependent. Although five children were still TD, whereas all parents had reached transfusion independence with age, outcome was not found to be different within generations (p= 0.16, Yates corrected χ2 test, df 1).

Disease evolution.

Two hundred twenty-two patients were available for long-term evaluation with a median follow-up of 111.5 mo and a total cumulative follow up of 2716 patient years. Figure 2 summarizes the evolution of the disease. Positive response to the first steroid treatment was obtained in 62.6% of the patients The response rate after the first steroid treatment was 64.7% with prednisone, 69.2% with prednisolone, and 75.5% with other steroids (p= 0.8), indicating that the type of steroid used for this first treatment did not influence the response rate. Steroid treatment had to be discontinued in 15.8% of the responders because of a progressive loss of response or severe side effects. Complications due to steroid therapy were noted in 19.7% of the patients under long-term therapy: osseous complications in 11 patients (5.6%), hypertension in three, diabetes mellitus in four, and growth retardation, presumably aggravated by steroid therapy, in 25 (13%). One infection-related death occurred in the group of SR, a 17-mo-old boy on long-term steroid therapy presenting with a silent peritonitis.

Figure 2
figure 2

Evolution of DBA cases, assessed after the first steroid course and at long-term follow-up.

Full size image

Among the 80 initial steroid nonresponders, eight (10%) could be weaned from transfusions during disease evolution, whereas four others were weaned from any treatment and 22 initially steroid-responsive patients became TD. Long-term follow-up showed a severe evolution in 40.5% of the patients: 65 patients (29.2%) required regular transfusion therapy, 13 (5.8%) required BMT, and 14 died (6.3%) (including two after BMT).

Complications due to transfusion therapy were observed in 69% of the long-term TD patients: iron overload despite regular subcutaneous desferrioxamine therapy occurred in 52 patients (62%), resulting in heart involvement in 12 patients (14.2%), liver dysfunction in 42 (50%), and endocrine deficiency in 21 (25%). Prevalence of hemochromatosis-related complication was significantly higher in TD DBA patients after age 10 compared with younger patients (71 versus 46%, p= 0.02), whereas no significant difference was noted among age-related groups when the cutoff point was set at age 15 (65 versus 54%, p= 0.3). Hepatitis C chronic infection was found in 6.6% of the transfused patients, all but one of whom was born before 1983. Three TD DBA patients were positive for both hepatitis C and HBsAg. Data on hepatitis B immunization were available for 178 patients. Immunization had been performed in only 59% of the TD patients (37.5% in non-TD patients). No significant difference in rate of immunization was observed among age groups (data not shown). No infection due to HIV was noted, nor any Creutzfeldt-Jacob disease. Transfusions and steroid treatment both could be discontinued in 46 of 222 patients (20.7%): 42 of 139 initially SR (30.2%) and four of 80 initially TD patients (5%), respectively. Among patients free of treatment, macrocytosis was observed in 33%, elevation of eADA in 80%, and increase in fetal Hb in 26%.

BMT was performed using HLA identical sibling donors (11 out of 13 patients), and donors were mismatched in two other patients. Overall results were satisfactory: survival, 84.6%; median follow-up, 36 mo (range, 3 to 138 mo) (Table 2). BMT-related deaths occurred in the two oldest patients at the time BMT was performed (15 y 11 mo and 18 y 4 mo, respectively).

Table 2 Characteristics and outcome of DBA patients treated with BMT * Cause of death, Aspergillus fumigatus pulmonary infection; § cause of death, graft versus host disease; M, male; F, female. Age at BMT or at last follow up is expressed in years and fractions of year.
Full size table

Malignancies occurred in three patients, two of whom were TD: femoral osteosarcoma in a 12-y-old girl, Hodgkin's disease in a 23-y-old man, and breast cancer in a 26-y-old transfusion-independent female. Fatal outcome occurred in 15 DBA cases (median age, 17 y 5 mo; range, 1 y 1 mo to 53 y 9 mo) (Table 3). Underlying conditions causing death among 14 of these patients were hemochromatosis in five, severe pancytopenia related to secondary chronic bone marrow failure in three, infections in three, cancer in two (Hodgkin's disease, osteosarcoma), and noninfectious complications of BMT (severe graft versus host disease) in one. More than one cause of death was reported in some patients; for instance, iron overload complications were often associated with infectious complications. Overall survival at 250 mo of age (20.8 y) was 88.6 ± 3.7% (mean ± SD) (Fig. 3). Survival at 250 mo was significantly better among those who initially responded to steroids (95.4 ± 2.7%,) compared with nonresponders (84.6 ± 10.8%) (p= 0.01, log-rank test).

Table 3 Clinical features and cause of death in DBA M, male; F, female. Response to first steroid course: 0, no response; 2, partial response with a high threshold in dose of steroid dependence; 3, transfusion independence. GVHD, graft vs host disease.
Full size table
Figure 3
figure 3

Actuarial survival calculated for patients born after 1977 reached 88:6 ± 3.7% (mean ± standard) at 250 mo (Kaplan-Meier technique).

Full size image

Prognostic factors.

Univariate analysis showed a significant association between absence of response to initial steroid course and four variables at diagnosis: young age at presentation, thrombocytosis above 450 × 109/L at diagnosis, absence of familial history, and presence of heart malformations. Logistic regression was performed with these variables included in the model. Three remained independently included in the model after logistic regression was performed (n= 163, r= 0.59) and were associated with absence of response to the first steroid course: young age at presentation, thrombocytosis at diagnosis, and absence of familial history. Similarly, six prognostic factors tested in univariate analysis against long-term outcome were significantly associated with unfavorable prognosis: neck malformations, thumb malformations, thrombocytosis at diagnosis, young age at presentation, premature birth, and absence of familial history. Among them, young age at presentation, absence of familial history, and premature birth remained the only independently associated variables with unfavorable outcome (n= 130, r= 0.84).

Epidemiology.

Among the 151 patients born in France or in overseas territories, 72 were born between January 1, 1983 and December 31, 1995. The annual incidence of DBA in France remained stable over this 13-y period with 7.3 new cases per million live births. No monthly variation of birth rate of DBA patients was observed in the whole population studied. A slight increase in prevalence of malformations was noted in children born at the end of the summer or in the fall (54.6% for the period September–November, p= 0.02; 50% for the period August–October, p= 0.04). However, these data did not reach the level of significance that was expected for this part of the study (p< 0.01). Further, this variation was not confirmed when the same data were analyzed on a monthly basis. Similarly, prevalence of DBA cases and malformations according to the month of conception did not show statistically any significant difference on a monthly or a 3-mo-period basis (data not shown).

DISCUSSION

The present study highlights the main features of the epidemiology and outcome of DBA through the analysis of a large well-characterized group of patients. Short-term prognosis is found to be dependent on steroid sensitiveness (62.6% of patients), whereas long-term prognosis is influenced by the morbidity encountered with transfusion programs (62% of TD patients) and mortality (18% of TD patients). Interestingly, long-term prognosis is statistically worse in sporadic cases compared with familial cases and in individuals with early onset of the disease, as previously reported (4). Analysis of our data enabled us to define the annual incidence of DBA in France to be 7.3 cases per million live births, without any effect of seasonality.

Presentation.

Our data show that short stature at presentation is significantly associated with malformative syndrome (5) and cannot be explained by other variables such as term of pregnancy or early onset of anemia. We were also able to confirm the variety of malformations and their frequency, as reported in earlier studies (3–5). The occurrence of malformative features among hematologically normal relatives of DBA patients remains a rare situation. Specifically, we observed three healthy relatives in a dominant DBA family who exhibited severe malformation of one hand and one of the two nonaffected children of a DBA mother who presented with Pierre Robin syndrome. Short stature at presentation (before any steroid therapy) seemed to be a true part of the malformative syndrome (4, 5) and raises the question whether or not short stature at adult age could be prevented. Postnatal onset of anemia observed in most of the DBA patients implies that erythropoiesis in utero is adequate in most cases for survival of the fetus. However, the prevalence of prematurity and intrauterine growth retardation is increased in DBA newborns compared with other newborns (19.9 versus 6% and 28.4 versus 10% in France, respectively, p< 0.0001). Malformations are not more frequently encountered in premature than in full-term birth DBA patients (46.8 versus 43.9%, respectively, p= 0.76) and thus do not seem to significantly contribute to premature birth in DBA. CBC was not always available at diagnosis or before transfusion, as initial data were often missing in the records of older patients and, in some patients, severe anemia at birth was followed by a prompt transfusion.

Disease evolution.

Our data also provided new insights into the therapeutic response and outcome in DBA patients. Initial steroid responsiveness is achieved in 61% of the cases. Long-term follow-up highlights time-related variations in response to treatments. Of the patients, 21.8% were finally weaned from any form of treatment. This group was composed mainly of SR (42 of 139, 30.2%) but also included rare individuals who were initially TD (four of 98, 4%). Thus, steroid response/resistance cannot be considered a definitive feature for a patient, as several patients experienced modifications in their status during long-term follow-up (13.5%). Specifically, 22 of 137 patients initially steroid sensitive became steroid resistant, whereas eight of 84 patients initially TD became SR. In some cases, response to steroid was achieved only after several steroid courses, suggesting that repeated steroid courses within the first decade of life should be considered in initially TD patients and are mandatory before major therapeutic decisions such as high-dose methyl prednisolone therapy, initiation of chelation, or BMT (15–20) are considered. Our study, however, does not provide data on the optimal number of steroid courses to perform in initially TD patients, as it was not previously considered a consensual treatment practice. Comprehension of such changes in therapeutic response would be a major step in understanding the pathophysiology of DBA. We also did not observe consistent loss of responsiveness to steroid therapy after interruption of the first course of steroid therapy (9.8% of initially SR and 17.8% of initially steroid-resistant patients changed status, respectively). These results do not support the hypothesis that an interruption of steroid therapy could be followed by a significant reduction in steroid response and favor the suggestion of transient discontinuation of steroid therapy and limited periods of transfusion support to promote catch-up periods for growth in steroid-responsive DBA patients with short stature. The clinical evolution we found in DBA patients after BMT emphasizes the overall good prognosis of this procedure as long as an HLA identical donor is available. The apparent discrepancy between outcome observed in the present study and that reported in the registry of the United States may be explained by the lower prevalence of HLA identical sibling donors (six out of 12 BMT) involved in the registry of the United States. In this earlier report, four out of five fatalities occurred in unrelated or mismatched transplants (6). Unfortunately, our data analysis has not been helpful in defining early and precisely the subset of patients who will remain TD and thus be potential candidates for early BMT. Deciding the optimal time for BMT in TD patients with an available sibling donor is a most distressing problem, as early BMT might be performed in patients who would still have a chance to reach transfusion independence, whereas, if BMT were performed later, transfusion-related complications might compromise the outcome. Yet HLA typing has to be performed during the first years of life in such situations, and cord blood of subsequent pregnancies should be systematically preserved (21).

No actuarial survival prediction can be made for the whole population of DBA patients from such a study design, and we proposed survival data for patients born since 1978. Janov et al. (4) carried out the first detailed study of a cohort of patients in a single clinical center and found a median survival of 31.2 y for patients born since 1960. Our data are consistent with this result, although most of the patients included in our study had not yet reached this age.

The major concern for TD DBA patients remains the complications related to this supportive therapy. Among our patients, 6.6% experienced blood-borne viral infections, and 63% exhibited iron overload-related complications despite regular chelation therapy, which was regularly performed in every TD patient. We did not observe any trend toward reduction in prevalence of iron overload complications among age groups, suggesting difference in compliance. Compliance to desferrioxamine therapy would need to be specifically assessed to further investigate this phenomenon. Importantly, immunization against hepatitis B should be more systematically performed in DBA, as all groups are potentially a population at risk for blood-borne infections.

Occurrence of malignancies in our series deserves further attention because other groups have reported an increased risk for malignancies in this condition (4, 22). The four patients reported by Janov et al. (4) exhibited acute myeloid leukemia after a prolonged period (15 to 30 y) and died between the ages of 24 and 43. Others have reported nonhematologic malignancies, as previously reviewed (22).

Epidemiology.

Early epidemiologic surveys of DBA were based on retrospective studies, leading to an underestimation of the incidence of this disorder [one case per million live births (23)]. Our data showing an incidence of 7.3 cases per million live births in France are in agreement with data from the United Kingdom (five per million), the Netherlands (four per million) (24), and Scandinavian countries [10 per million (L. Wranne, personal communication)]. Although the data on incidence can be considered reliable, disease prevalence data are certainly underestimated because most steroid and transfusion-independent adult patients are lost to follow up after childhood. Inclusion of patients older than 20 y in databases relies usually on new contacts of these individuals with a hematology or internal medicine group because of relapse of anemia, genetic counseling, pregnancy, the birth of an affected child, the occurrence of a malignancy, or through the family associations. Our data do not support the hypothesis of a seasonal factor previously suggested by Ball et al. (5). Indeed, we did not find any significant seasonal variation in incidence of DBA or any significant prevalence of dysmorphic features according to birth or conception date. The moderate increase in prevalence of malformations at the end of the summer or during the fall can be considered nonsignificant due to the large number of statistical tests performed. No gender difference was observed in phenotypic expression of the disease, and evaluation of familial cases did not suggest the existence of an effect of parental imprinting or any anticipation phenomenon in dominant cases.

The present study as well as a number of earlier studies implies that DBA is a heterogeneous clinical condition with regard to existence of malformations, sporadic cases or familial history, and response to steroid therapy. The novelty of our study is that we have been able to document these diverse features as statistically linked, giving strong support to the hypothesis that a common underlying genetic defect or mechanism could explain most of the variations in phenotypic expression of the disorder. To establish the diagnosis of DBA, Fanconi's anemia and parvovirus B19 must be ruled out, as therapeutic options differ (7). Steroid response/resistance does not seem to be a permanent feature in DBA patients, as 13.5% of DBA patients experience modifications in their status. Further, the occurrence of complication during steroid treatment or the existence of a threshold above 0.5 mg·kg−1·d−1 often requires a long-term transfusion program. Repeated attempts of steroid treatment, however, are useful, as they may reveal an improved steroid sensitiveness. Further studies on growth will be needed, as it seemed that growth failure related to steroid treatment requires a short-term transfusion program in some cases to allow for catch-up growth. The prognosis encountered by TD patients remains impaired by complications of iron overload and blood-borne infections. BMT has proven to be an effective treatment option when an intrafamilial genoidentical donor is available. Until results of further therapeutic trials are available, the prognosis for steroid-resistant patients will remain a significant concern. Recent research on genes underlying DBA may help in understanding the phenotypic variations of the disease as well as the mechanism(s) of steroid sensitiveness and, thus, lead to better treatment planning for DBA patients (25–27). At the time this study was completed, identification of mutations in a gene encoding the ribosomal protein S19 had opened interesting fields for research in understanding these features and, hopefully, therapeutic management (28). Genotypic analysis of the patients in the present study may explain some of the prognostic factors that we report here.