This retrospective–prospective study using real-world data to assess the evolution of bone disease in patients with GD1 being treated with velaglucerase alfa provided valuable insights into the impact of the treatment, and into the quality and effectiveness of patient monitoring in clinical practice in France. Despite the publication of a detailed protocol for the monitoring of bone disease in patients with GD, we found that the quality of bone MRI data collected in clinical practice was often insufficient to allow for semiquantitative assessments of treatment responses through calculation of BMB scores. However, the centralized qualitative assessment of real-world MRI data used in this study provided evidence of the positive impact of velaglucerase alfa on bone disease, with improvements in bone infiltration being observed in treatment-naive patients and stabilization of bone infiltration being observed in treatment-switched patients. Furthermore, reductions in acute and chronic bone pain, and improvements or stabilization of hematologic parameters, and visceral manifestations were observed, providing further evidence from clinical practice of the effectiveness of velaglucerase alfa in allowing patients to achieve and maintain the well-established goals for GD treatment.
Assessing the extent of bone marrow infiltration in patients with GD is essential for evaluating the extent of bone involvement, monitoring patient responses to ERT, and for guiding therapeutic decision making and optimizing treatment regimens [3, 39]. The BMB score, which provides an MRI-based semiquantitative evaluation of bone marrow infiltration based on the distribution of lesions and the change in signal intensity, is one of the most widely used methods in clinical studies [3, 40]. This method has the advantage of being more reflective of whole-body bone marrow involvement because it includes assessment of the spine and the femur, and is simpler to use and more widely accessible than Dixon quantitative chemical shift imaging (QCSI) assessments of the bone marrow fat fraction because it uses conventional MRI imaging [14, 17, 36, 39, 41]. However, the interobserver agreement of BMB scores has been questioned, even between experienced radiologists [42]. In addition, BMB scoring has been found to be less reliable for assessing bone infiltration in younger patients with GD, due to potential masking of the true extent of bone infiltration by the higher proportion of red bone marrow normally present in children and young adults [3, 43, 44]. The findings from the current study have highlighted the problems associated with the use of the BMB scoring for monitoring bone involvement in GD in real-world clinical practice. Longitudinal assessments of BMB scores rely on the collection of high-quality sequential MRI data according to a rather strict protocol to ensure consistency between device settings and scanning parameters during follow up. The MRIs used in this study were conducted by local radiologists using a range of MRI machine models from multiple centers and varied scanning parameters, resulting in large technical variations in the quality of the images obtained. Despite the availability of detailed procedures for conducting bone MRIs in patients with GD [35], the centralized analysis of the MRI data revealed that in many cases these good practice guidelines were not followed, particularly the recommendations to collect image sequences of both the distal and proximal femur and to acquire T1- and T2-weighted sequences without fat-suppressed imaging to allow optimal comparison of the relative changes in signal intensity between healthy and diseased bone areas. These findings indicate that more specialized training needs to be provided to local radiologists on the acquisition and interpretation of data for calculating BMB scores. In countries where resources permit, this may need to be combined with centralized analysis through a national reference platform to limit interobserver variability in BMB scoring, or alternatively, the development of semi-automated techniques could be investigated to improve consistency.
As a result of the technical limitations associated with data acquisition, only one of the 17 adult patients included in the current study had interpretable data for femur BMB scoring, and only seven patients had interpretable data for spine BMB scoring. This absence of adequate data for the assessment of total BMB score led us to explore whether a subjective qualitive assessment of the change in bone infiltration (classified as worsened, improved or stabilized) would provide a more feasible method for analyzing the response to velaglucerase alfa treatment in clinical practice. The results of the centralized analysis indicated that this qualitative method could indeed be used to monitor changes in bone infiltration in the spine and femur over time, and to identify statistically significant differences between treatment groups. In addition, qualitative analysis by the centralized radiologist also appeared to provide valuable information about the change in bone infiltration in the three children included in the study cohort. Although the normal developmental changes in red marrow made it more challenging to assess improvements in bone infiltration in these younger patients compared to in the adults with GD, the masking effect of bone marrow maturation was not considered to have prevented the visualization of significant worsening of bone infiltration during qualitative assessment. Thus, although larger validation studies are needed, our findings suggest that the qualitative bone infiltration analysis used in this study could provide an alternative, less stringent, and easier-to-interpret method that could be used by local radiologists to assess bone disease in patients with GD, particularly in countries or regions where medical resources are scarce and access to GD expert radiologists is extremely limited.
The results of the qualitative analysis provided valuable real-world insights into the impact of velaglucerase alfa treatment on bone involvement in patients with GD, with treatment-naive patients and those with shorter treatment durations (3.5 years on average) tending to show improvements in femur and spine infiltration, and patients who switched treatments and treatment-naive patients with longer treatment durations (6 years on average) generally having stable bone disease. Importantly, none of the patients showed signs of worsening bone disease during velaglucerase alfa treatment. Our results are therefore consistent with findings of previous clinical studies in which assessment of BMB scores showed that velaglucerase alfa treatment led to a significant reduction in bone infiltration [14], with largest reductions being observed during the first 5 years of ERT, followed by long-term stabilization after 5 years [16, 17]. As pointed out in these studies [16, 17], the stabilization of bone infiltration after 5 years of treatment suggests the need to revise current bone MRI monitoring protocols, perhaps increasing the interval between MRI evaluations for bone infiltration in patients who are adherent to treatment and have stable bone disease. Indeed, the current PNDS guidelines recommend bone MRI at treatment initiation, after 1 year, and then every 2 years once the disease has stabilized [28]. Improving adherence to the PNDS guidelines and ensuring that patients are monitored by MRI, particularly when initiating ERT or switching between therapies, is essential to allow for meaningful evaluation of treatment responses and for evidence-based updates of current monitoring recommendations.
In addition to improving bone infiltration, reducing bone pain is one of the key therapeutic goals in patients with GD [26, 28]. Our analysis of clinical medical records indicated that this goal was met in our cohort: no new cases of acute bone pain were reported during velaglucerase alfa treatment and the resolution of acute bone pain was reported in two patients. Although available data on pain severity were limited, the velaglucerase alfa treatment also seemed to lead to a reduction or stabilization of chronic bone pain. Similar findings have been reported previously in several studies examining the impact of the alternative ERT, imiglucerase, on bone pain in patients with GD [9, 13, 18].
Based on the analysis of bone imaging records written by local radiologists, the large majority of patients in our cohort had existing bone lesions before the initiation of velaglucerase alfa therapy. As noted in previous studies, ERT cannot reverse all of the existing bone manifestations of GD, most notably avascular osteonecrosis and disease-related complications such as secondary osteoarthritis and fracture deformities [3], and thus the goal of therapy is to prevent bone infiltration and the occurrence of new lesions [26, 28]. Velaglucerase alfa treatment appeared to allow all of the patients included in our study to achieve this goal. However, the occurrence of bone lesions in patients receiving ERT has been reported in previous studies [45–48], particularly in patients that initiated ERT more than 2 years after GD diagnosis [49]. Indeed, in our study the mean time between diagnosis and initiation of any form of GD treatment was 2.6 years in treatment-naive patients and 9.9 years in patients switching treatments. Thus, the early detection of new lesions and surveillance of the severity of existing lesions remains essential for patients receiving ERT, not only for guiding the adjustment and optimization of velaglucerase alfa treatment, but also to allow timely intervention with supportive therapies and interventions such as prosthetic replacement to maintain or improve patient mobility and quality of life. In addition, other bone complications, not specifically related to GD were reported in over 50% of the patients in our study. Clearly, there is a need to ensure that these nonspecific bone lesions are not overlooked during monitoring of the complex bone manifestations of GD and that correct diagnosis and appropriate management are provided.
Hepatomegaly and splenomegaly are hallmark manifestations of GD1, present in between 60% and 90% and more than 90% of cases, respectively [1]. The PNDS recommends regular monitoring of liver and spleen volumes every 6 months during the first year of treatment and then biannually after stabilization of organ volumes, by either abdominal MRI or ultrasound [28]. Although ultrasound has the advantage of being more accessible and affordable than other imaging modalities, it provides a less comprehensive assessment of organ involvement than MRI [44]. In contrast, MRI data can be analyzed using semi-automated methods for measuring organ volumes, improving measurement accuracy and reproducibility [50]. In our study, less than half of the patients had available abdominal MRI data collected around the time of initiation of velaglucerase alfa treatment and only around half of the patients had longitudinal abdominal MRI data. However, contrary to the bone MRIs, all of the abdominal MRIs conducted provided interpretable, although not always optimal, cross-sectional data for the centralized analysis of organ volumes, demonstrating that the semi-automated method used in this study was sufficiently robust to overcome the variations in scanning parameters and sequence types associated with MRI data collected in clinical practice. Thus, while clinical examination and ultrasound maybe sufficient for routine long-term management when MRI facilities are scarce, when MRI is widely available, this technique could be used in clinical practice to monitor visceral disease severity and treatment responses.
The treatment goals for the visceral complications in GD1 are to reduce (within the first two years of treatment) and then stabilize organ volumes [28], ideally to within less than 1.0 to 1.5 MN for the liver and 2 to 8 MN for the spleen [26]. The centralized analysis of liver and spleen volumes demonstrated that these goals were met in our patient cohort, with treatment-naive patients showing decreases in liver and spleen volumes by the last visit to achieve average volumes of 1.2 MN and 3.9 MN, respectively, and treatment-switched patients showing stabilization of organ volumes (0.8 MN for the liver and 2.6 MN for the spleen at the last visit). These findings are therefore consistent with those of previous studies showing that velaglucerase alfa treatment leads to near-normalization of hepatomegaly and major reductions in splenomegaly within around 4 years of initiating treatment [12, 51, 52].
The same pattern of improvement in treatment-naive patients and stabilization in patients switching to velaglucerase alfa was observed for hemoglobin concentrations and platelet counts during the study. All patients had normalization or stabilization of hemoglobin and platelet counts at the last visit. Such normalization and maintenance of hematologic parameters in the 4 years after initiating treatment has been reported previously in patients receiving velaglucerase alfa [12], and has often been observed within the first 2 years post treatment initiation [51]. These findings are consistent with those of previous studies clearly demonstrating the effectiveness of velaglucerase alfa treatment in allowing patients to achieve the therapeutic goals of preventing anemia and reducing bleeding tendency, as well as the complications related to these hematologic manifestations [26, 28, 52]. Finally, monitoring of chitotriosidase activity, a well-established biomarker of GD severity and treatment responses [47, 53, 54], revealed decreases in activity both in the treatment-naive patients and treatment-switched patients.
The real-world ambispective design of the current study allowed the long-term impact of velaglucerase alfa treatment in GD1 to be assessed in clinical practice, in a patient population that was homogeneous in terms of the dose and frequency of velaglucerase alfa treatment received, and in both treatment-naive patients and those who had switched treatment. However, this real-world approach led to several study limitations. First, the amount and quality of the MRI data collected in clinical practice were insufficient to allow assessment of the primary study endpoint (i.e., the change in BMB scores). However, this lack of available data led us to explore the potential of an alternative and less stringent qualitative method for assessing bone marrow involvement. Future studies would allow us to further evaluate the suitability of this method for monitoring the bone marrow response to ERT in real-world clinical practice, and to more closely examine the suitability of the method for use in younger patients. Second, the size of study population was small and limited the power of the study to detect statistically significant between-visit differences and between-parameter correlations for some measures, most notably in the occurrence of bone pain. The small size of the study population was associated with several factors. GD is a rare disease with an estimated prevalence of 1 in 140 000 in France [29]. According to the CETG registry, there were 97 patients with GD living in France who had received at least one dose of velaglucerase alfa at the time of the study. Among the centers managing these patients, only those treating more than one patient were invited to participate and only patients with GD1 and digital records of MRIs conducted within the 5 years preceding velaglucerase alfa initiation, or within the 3 months following treatment initiation, were eligible for study inclusion. Future studies involving larger populations would help to further clarify the extent to which ERT can prevent bone infiltration, and allow more detailed characterization of patients who are at highest risk of bone disease progression despite treatment. Ideally, a more organized, well-funded, international approach is required, perhaps using a purpose-designed platform for data collection. However, to allow pooling of all the collected data it is important that consensus is reached on MRI monitoring protocols and on the terminology used to describe the bone lesions, with the terms osteonecrosis, avascular necrosis, aseptic osteonecrosis and bone infarct often being used interchangeably in the literature, regardless of the anatomical location of the lesion. The size of the population also did not allow for a separate evaluation of disease evolution in patients that had undergone splenectomy, although the impact of this intervention on treatment responses has been reported previously [12]. Due to the retrospective nature of part of the study, the only GD marker with a sufficient amount of data available for longitudinal analysis was chitotriosidase, as more recently validated markers, such as glucosylsphingosine (lyso-Gb1) [55], were not commonly used in clinical practice at time when many of the patients included in study initiated velaglucerase alfa treatment. Finally, the duration of the interval between the first and last available measure varied for each variable, reflecting the monitoring regimen of individual patients.