The French Gaucher’s disease registry: clinical characteristics, complications and treatment of 562 patients

Background Clinical features, complications and treatments of Gaucher’s disease (GD), a rare autosomal–recessive disorder due to a confirmed lysosomal enzyme (glucocerebrosidase) deficiency, are described. Methods All patients with known GD, living in France, with ≥1 consultations (1980–2010), were included in the French GD registry, yielding the following 4 groups: the entire cohort, with clinical description; and its subgroups: patients with ≥1 follow-up visits, to investigate complications; recently followed (2009–2010) patients; and patients treated during 2009–2010, to examine complications before and during treatment. Data are expressed as medians (range) for continuous variables and numbers (%) for categorical variables. Results Among the 562 registry patients, 265 (49.6%) were females; 454 (85.0%) had type 1, 22 (4.1%) type 2, 37 (6.9%) perinatal–lethal type and 21 (3.9%) type 3. Median ages at first GD symptoms and diagnosis, respectively, were 15 (0–77) and 22 (0–84) years for all types. The first symptom diagnosing GD was splenomegaly and/or thrombocytopenia (37.6% and 26.3%, respectively). Bone-marrow aspiration and/or biopsy yielded the diagnosis for 54.7% of the patients, with enzyme deficiency confirming GD for all patients. Birth incidence rate was estimated at 1/50,000 and prevalence at 1/136,000. For the 378 followed patients, median follow-up was 16.2 (0.1–67.6) years. Major clinical complications were bone events (BE; avascular necrosis, bone infarct or pathological fracture) for 109 patients, splenectomy for 104, and Parkinson’s disease for 14; 38 patients died (neurological complications for 15 type-2 and 3 type-3 patients, GD complications for 11 type-1 and another disease for 9 type-1 patients). Forty-six had monoclonal gammopathy. Among 283 recently followed patients, 36 were untreated and 247 had been treated during 2009–2010; 216 patients received treatment in December 2010 (126 with imiglucerase, 45 velaglucerase, 24 taliglucerase, 21 miglustat). BE occurred before (130 in 67 patients) and under treatment (60 in 41 patients) with respective estimated frequencies (95% CI) of first BE at 10 years of 20.3% (14.1%–26.5%) and 19.8% (13.5%–26.1%). Conclusion This registry enabled the epidemiological description of GD in France and showed that BE occur even during treatment.

Enzyme-replacement therapy (ERT; alglucerase [Ceredase©, Genzyme Corporation, available since 1991] [18], followed by imiglucerase [Cerezyme©, Genzyme Corporation, available since 1996], velaglucerase [Vpriv©, Shire, available since 2010] [19], and taliglucerase [Uplyso©, Pfizer, only authorized for temporary use] [20]), is the reference treatment. Substrate-reduction therapy (SRT), namely miglustat (Zavesca©, Actelion, available since 2002) [21], is indicated for moderate GD when ERT is unsuitable. In June 2009, an acute imiglucerase shortage occurred because of viral contamination (Vesivirus 2117) of cell cultures and other production problems [22]. Since then, that ERT has been in short supply, which was further aggravated in August 2009. Genzyme Corporation developed an international registry [23] and several countries, e.g., Spain [24,25], Brazil [26] or Japan [27], identified GD cohorts and established exhaustive national registries. While the international registry conducted many important ancillary studies [28][29][30][31][32][33][34], its non-exhaustive cohort did not address public health issues in terms of incidence, prevalence and monitoring of care of GD patients. Since 2004, France has created referral centers dedicated to the clinical management of rare diseases, and assigned them several objectives, e.g., improving overall patient clinical care and professional practices, and collecting epidemiological data. In this context, a Referral Center for Lysosomal Diseases (RCLD) was established and a national GD-patient registry was created, in 2009, as a means to examine and meet some of those goals.
The main aims of this study were to describe the epidemiological profile of GD patients in France: GD demographic, clinical, biological and genetic features; complications in patients with follow-up (2009-2010); and treatments for those with recent (2009-2010) follow-up based on data collected since 1980 and available in the French Gaucher Disease Registry (FGDR).

Registry design
The FGDR developed and maintains a designated RCLD since 2009. Its Evaluation of Gaucher-Disease-Treatment Committee (EGDTC) is a national scientific committee to monitor and optimize GD management in France. The French Data-Protection Commission's (CNIL) approval of the FGDR required oral or written informed consent from patients or their parents. Data from patients who did not consent were not entered. The FGDR was finally certified in 2009 by the French Institute for Public Health Surveillance (InVS) and the French National Institute of Health and Medical Research (INSERM). All GD patients living in France and having ≥1 consultations (i.e., hospitalization or outpatient consultation with a GD specialist) since 1980 were included. For all patients, GD was diagnosed by demonstration of deficient glucocerebrosidase activity in leukocytes [35] or cultured skin fibroblasts. Exhaustive identification of cases was achieved through 3 sources. Only 3 diagnostic laboratories (all included in EGDTC) are accredited in France and, therefore, identify all patients with an enzymatic assay for GD. Based on our Reference Center's expertise, the French national health insurance (Rare Disease Committee with EGDTC members) validates each GD diagnosis and authorizes coverage for its treatment. Once experts have validated the case, they can ask the treating physician to include the patient in our registry. The indications for treatment are well established to maximize efficacy and avoid unnecessary health insurance expenditures. Each treating physician contacted allowed access to the medical data entered in the FGDR, which is certified by InVS and INSERM.
Each patient's data were also collected by RCLD physicians or clinical research assistants. The FGDR director controlled data quality. Dr D. Hamroun developed the original Internet software for the FGDR, using 4 th Dimension language from 4D (www.4D.com). Data were collected retrospectively between 2009 and 2010, and as of 2011, all data have been recorded prospectively.
A standardized case-report form was used to collect the following information: initial data (age at diagnosis, sex, history related or unrelated to GD, symptoms leading to diagnosis and first symptoms, first diagnostic exam, phenotype, genotype and affected family members); clinical information during the first consultation, at diagnosis and throughout follow-up; body mass index expressed according to the World Health Organization classification; organomegaly (liver and/or spleen, ultrasound measurement of the largest diameter); biological findings initially and throughout follow-up (hemoglobin level, platelet count, leukocyte count, chitotriosidase, ferritin, ACE, TRAP, gammaglobulin (with respective normal values of >12 g/dL, >150×10 3 /mm 3 , >4 × 10 3 /mm 3 , <100 nmol/mL/h, <250 ng/L, <45 IU/L, <7 IU/L and <13 g/L). Plasma chitotriosidase activity was determined using the fluorescent substrate 4-methylumbelliferylβ-D-N,N′,N′′-triacetylchitotriose [10]; ACE, TRAP, ferritin and other markers were measured in the appropriate local laboratories. Bone findings (X-rays, magnetic resonance imaging and, for some patients, scintigraphy and dual-energy X-ray absorptiometry) were recorded during follow-up, with identification of intercurrent events, particularly bone complications. Bone events (BE) were defined clinically, using the bone indications for treatment recommended by the French National Health Authority [36]: avascular necrosis of an epiphysis, bone infarct, pathological and/or vertebral compression fracture(s). Each BE caused a clinical manifestation and was confirmed radiologically. Bone pain alone was not considered a BE without radiological confirmation. Acute bone pain defined a bone crisis. Bone crisis was included in BE only when a bone infarct was identified. Any event, GD-related or not, occurring during followup and monitoring of GD-specific therapy was also recorded.

Study design
This investigation was undertaken to describe and analyze clinical, biological, radiological and therapeutic data recorded in the FGDR for all patients from diagnosis until 31 December 2010, the closing date. The local Institutional Review Board of Northern Paris Hospitals, Paris-Diderot University, AP-HP (Ethics Committee) reviewed and approved the research project.
To simplify the description, we defined 4 groups: the entire cohort and its subgroups. Data from the entire cohort of patients entered in the FGDR described, when available, diagnosis characteristics for these patients, the GD-incidence rate (defined as total number of cases diagnosed between 1980 and 2010 divided by the total French population during the same period), birth incidence rate (defined as the total number of cases diagnosed between 1980 and 2010, divided by the total number of live births during the same period) and prevalence were estimated for the French population. For patients with ≥1 follow-up visits in addition to the initial assessment form, we investigated their GD complications (splenectomy, Parkinson's disease (PD), monoclonal gammopathy (MG), BE, first treatment and deaths). Recently followed patients had consulted in 2009-2010: a map showing the locations of hospitals monitoring them was drawn. For patients seen and treated in 2009-2010, BE were analyzed, before and under (ERT and/or SRT (ERT/SRT)). Clinical, biological and radiological monitoring of these recently treated patients was also investigated. Specific GD ERT (imiglucerase, velaglucerase, taliglucerase, miglustat) was studied, particularly during the period of imiglucerase shortage (June 2009-December 2010). Patients were distinguished according to their age on 1 June 2009 (≤15 or >15 years) for the description of the shortage that began at that time. That age was chosen because it defines the limit between adult medicine and pediatrics.

Statistical analyses
All statistical analyses were computed with SAS software (version 9.2; SAS Institute Inc; Cary, NC). Data are expressed as medians ((range) or interquartile range [IQR; Q1;Q3]) for continuous variables and numbers (%) for categorical variables. Because this was a retrospective study, some data were missing, particularly at the onset of follow-up (during the diagnosis phase) or at treatment onset. Given the demonstrated relatively stable clinical and laboratory parameters of untreated patients after GD diagnosis [37], the biological data during the next 2 years changed only minimally and were considered similar to those at diagnosis. Likewise, data for the previous 2 years under ERT/SRT were stable compared to those at treatment onset. Under treatment data were the last values before the end of therapy or at the closing date. When ERT/SRT was interrupted for <6 months, patients were always considered to be on treatment.
Non-parametric tests were used to compare categorical variables (Fisher's exact test) across patient subgroups. A two-sided p<0.05 was considered significant. For recently treated patients, time to first BE was estimated with the Kaplan-Meier method for 2 periods: diagnosis to treatment onset (before ERT) and first ERT to closing date (under ERT), with only the first BE occurring during each period being considered. Data were censored when no BE occurred prior to ERT start for the first analysis, and until the closing date or treatment discontinuation for the second. We aimed to study a risk effect of BE. First, the impact of splenectomy, time to treatment onset (< or ≥2 years after diagnosis, based on Mistry et al.'s demonstration of lower BE risk after the latter [38]) or age at diagnosis (≤ or >15 years) on BE occurrence was tested using the log-rank test; second a Cox model was used to derive a predictive model. When several univariate model covariates were significant, a multivariate Cox model with backward selection was used to retain only significant ones. For BE under ERT, the impact of BE occurrence before treatment was also tested.

Followed patients
A total of 378 patients, predominantly type 1, had ≥1 follow-up visits after their initial evaluations. The median follow-up duration was 16.2 (0.1-67.6) years. Their characteristics at diagnosis are reported in Table 1.
During follow-up, 225 complained of chronic bone pain or clinical bone crisis, 231 had splenomegaly and 163 had hepatomegaly at least once.

Discussion
To date, no other publication has analyzed the comprehensive data entered in the FGDR for 562 patients, minus 3 who refused to participate and 97 who died, leaving 465 patients (among 65.8 million inhabitants), yielding prevalence of 1/140,000 inhabitants in France, a number that is probably underestimated. Concerning the entire cohort, although type 1 predominated (85%), types 2 and type 3 represented 4% each, along with 37 (6.9%) perinatal-lethal type. Moreover, the type-2 incidence was the same as that of type 3 but its prevalence was low because of its associated early mortality. The recent publication on the exhaustive Spanish registry [26] reported data similar to ours, with 88.3% type 1, 6.7% type 2 and 5% type 3. Our birth incidence (1/50,000) was higher than previously reported for the GD frequency in non-Jewish populations from EU countries [25,39,40], with a prevalence (1/136,000), close to that of the Spanish registry (1/149,000) [26]. Bone-marrow aspiration (or biopsy) remained the most common laboratory test (57%) providing the GD Figure 3 Impact of BE before treatment on BE occurrence under ERT/SRT for 247 treated GD patients. The solid bold grey line represents patients without BE before treatment; the solid bold black line represents the times to first BE. Dashed lines represent the 95% CI of those curves; the estimated probability of BE occurrence after 10 years reported on the y-axis. diagnosis. It is usually the first-line analysis when thrombocytopenia is associated (or not) with splenomegaly and there is no reason to think of immune thrombocytopenia purpura. It is not mandatory and should not be done if the GD diagnosis has been established by enzymatic assay or is already strongly suspected (e.g., possible family history). Rarely, bonemarrow aspiration was considered "normal" but another sample contained the characteristic GD cells. Fourteen (3.7%) of our 378 followed patients had PD, reaching a prevalence comparable to that reported by Bultron et al. [41]. MG and polyclonal gamma globulinemia occur frequently in GD [42][43][44][45]. Among the 378 followed patients, 46 (12.2%) had MG, a rate within the previously reported range (1% [42] to 35% [45]), and median gamma globulinemia at ERT/SRT onset in recently treated patients was 21.7 g/L. Usually, MG is unaffected by ERT [43,44]. However, for patients whose MG was diagnosed under treatment, no pretreatment evaluation was available, and MG had probably been present at treatment onset.
Before 1991, splenectomy was the only available treatment but, since then, it should not have been performed (albeit with exceptions) as a GD treatment. However, it has been used sometimes as a diagnostic tool when splenomegaly and thrombocytopenia coexisted, but should no longer be. Fourteen splenectomies were done after 1991 and after GD diagnosis, usually for patients with splenic complications (splenic infarcts, spleen rupture or large fibrous splenomegaly not amenable to ERT) or a mistaken indication.
BE are the most serious GD complications. They are usually prevented by ERT/SRT, with substantial attenuation of bone pain, bone crises and bone-mineral density [46], although the BE decrease is difficult to evaluate without randomized placebo-controlled trials. In addition, the definition of BE is not homogeneous across studies. Apparently, ERT/SRT does not prevent all BE, as indicated by the estimated respective probabilities of BE occurring by 10 years before and during treatment of 20.3% and 19.8%. It is likely that patients on ERT/SRT would probably have had more complications had they not been treated. Furthermore, we showed that BE before treatment increased the risk of BE under ERT/SRT and was the only factor retained in our multivariate analysis. Note that, as reported by Mistry et al. [38], our univariate analyses also found splenectomy and treatment >2 years after GD diagnosis to increase that risk, while sex and age at diagnosis ≤15 years were associated with increased risk of BE before but not under ERT/ SRT. Thus, BE persist as a problem that is not fully resolved by treatment. The continuing challenges remain how to identify patients at risk before and under ERT/ SRT, and then to decide whether or not these patients would benefit from earlier treatment onset and/or dose intensification.
In summary, the FGDR strong points are its comprehensiveness, independence, accreditation and/or certification by the various health authorities and cooperation generated among the different French centers. This registry also had to manage the imiglucerase shortage, when more severe GD and children were accorded priority treatment. The FGDR also enabled, during that shortage, nationwide management of the ERT/SRT stock and selection of those patients most in need of therapy (velaglucerase and taliglucerase). In France, GD-patient management is organized so that patients receive treatment near their homes, which improves their quality of life. Even though monitoring is not centralized, the FGDR identification and tracking of patients should contribute to improving their specific care management.

Competing interests
Research grants from Genzyme France to University Paris-Diderot to fund statistical analyses and to AP-HP to finance data acquisition. Research grant from Shire France to APRIMI (Beaujon Hospital's association) to finance data acquisition. C. Serratrice, L. Rossi-Semerano and B. Grosbois, received consulting fees, speaking fees, and/or honoraria from Genzyme (less than $10,000). C. Serratrice, D. Heraoui, F. Camou, A. Masseau and B. Grosbois received consulting fees, speaking fees and/or honoraria from Actelion (less than $10,000). C. Serratrice, F. Camou and B. Grosbois received consulting fees, speaking fees, and/or honoraria from Shire (less than $10,000).
Authors' contributions JS, FM, MV, NB, OF and BF designed research analyzed and interpreted data. All authors, except for FM, MV and DH were involved in treating patients and collecting data. JS had full access to all of the study data and takes responsibility for their integrity and the accuracy of the data analysis. DH developed the original software for the FGDR. JS, MV, NB and FM wrote the draft of the paper, which was then corrected and approved by all authors. All authors read and approved the final manuscript.