Fibrodysplasia ossificans progressiva (FOP) is a rare genetic disease affecting 0.88 per million individuals worldwide (1). Slowly but steadily, bone is formed in connective tissues such as muscles, ligaments and tendons throughout the body, leading to progressive immobility and ultimately an early death (2). In 2006, it was discovered that a single nucleotide substitution encoding the activin receptor type 1 (ACVR1)/activin receptor-like kinase 2 (ALK2) was the cause of FOP, and further genetic analyses revealed that nearly 97% of FOP cases are caused by the same mutation (c.617G>A, R206H), although other mutations have been described (3). Despite the fact that most patients share the same mutation, severity and rate of disease progression vary greatly and can be difficult to predict (4).

Heterotopic ossification in FOP is thought to take place in two separate manners. Classically, heterotopic bone in FOP is formed during so-called “flare-ups”, i.e., acute episodes of painful soft tissue swelling triggered by a minor trauma or an infection, but sometimes also occurring spontaneously (5). The initial local inflammatory response in muscle tissue is followed by endochondral ossification of the muscles and connective tissues (6). Recently, when analysing a series of MRI and PET/CT scans over time, areas were identified where heterotopic bone was formed without any typical flare-up symptoms, suggesting that a lower grade type of bone formation can also take place in patients with FOP, which is unrelated to the flare-ups (7). This was also confirmed in an ongoing natural history study in FOP, in which it was noted that nearly half of the patients reported instances of new heterotopic ossification (HO) without having experienced a flare-up (8). Treatment for FOP focuses on preventing soft tissue injury that can provoke a flare-up. Anti-inflammatory medication is often used empirically to reduce further bone formation once a flare-up has started, although there is little scientific evidence for this approach. Nevertheless, several therapeutic targets are being explored in multiple ongoing clinical trials, which will hopefully lead to better treatment and management of this devastating disease (9, 10).

Multiple tools have been developed to assess disease activity and progression of FOP. Assessment of joint mobility by goniometer measurements or cumulative scores such as the CAJIS (cumulative analogue joint involvement scale) reflect overall mobility and disease burden of a patient with FOP over time (11), but only moderately (r = 0.57) reflects the amount of heterotopic bone being formed (2). In clinical practice, however, it usually is more common to determine disease activity at the time of assessment for clinical decision making, rather than that it is based on changes over time. Clinical symptoms of a flare-up, such as pain, swelling, erythema and warmth, are non-specific, and it is difficult to predict whether the acute phase of the flare-up will result in HO formation or whether it will resolve itself (2). A multitude of potential biomarkers involved in inflammatory, chondrogenic and osteogenic processes have been investigated in patients with FOP, both during and in the absence of flare-up activity, and although some were markedly elevated in patients with FOP, none have shown to be capable of consistently reflecting disease activity and adequately predicting HO formation (12, 13).

Imaging techniques can be of great value in characterising disease activity and progression in FOP. Computed tomography (CT) scans can measure heterotopic bone volume over time, making it possible to evaluate disease progression. MRI (magnetic resonance imaging) and ultrasonography are able to detect soft tissue oedema associated with the inflammatory stage of HO, but they are less useful for detecting early bone formation and for accurately quantifying bone volume. Positron emission tomography (PET) is emerging as a new promising tool in FOP, as [¹⁸F]sodium fluoride [¹⁸F]NaF) PET can detect bone formation before it is visible on conventional CT (14).

Several methods exist to quantify [¹⁸F]NaF kinetics. A compartment model together with non-linear regression (NLR), as proposed by Hawkins et al., is considered to be the most accurate method for quantification of [¹⁸F]NaF uptake, but is also the most complex one (15, 16). The standardized uptake value (SUV) provides a simpler method towards assessing [¹⁸F]NaF uptake at any given time. SUV is the ratio of the image derived tissue radioactivity concentration divided by the whole body concentration based on the known dose of injected radioactivity, normalised to an anthropomorphic factor such body weight (BW). Though SUV is relatively easy to measure, it is prone to bias due to changes in blood flow and cannot assess the full range of [¹⁸F]NaF kinetics. Use of SUV to measure [¹⁸F]NaF uptake has already been validated against full kinetic NLR analysis in normotopic bone formation (17). Whether this simpler method can also be used to evaluate heterotopic bone formation and metabolism in FOP is unknown. Uptake in FOP may be markedly different from other metabolic diseases, given the genetic mutation interfering in the osteogenic pathways. In addition, drug therapies aimed at altering these pathways as part of a treatment for FOP may affect the rate of bone metabolism and, therefore, [¹⁸F]NaF kinetics even further. Potentially even more problematic, is that instead of altering bone metabolism, these drugs may also alter perfusion which, in turn, could give misleading results. For example, if metabolism does not change after therapy, but blood flow changes with 50%, SUV will also show an (erroneous) change of about 50%.

To assess various quantitative parameters reflecting [¹⁸F]NaF uptake in FOP, dynamic [¹⁸F]NaF PET/CT scans were performed in seven FOP patients who participated in the LUMINA-1 trial, evaluating [¹⁸F]NaF kinetics at baseline (T₀) and after one year (T₁) when all patients were receiving the trial drug garetosmab (an anti-activin A antibody).

Quantification of ¹⁸F-fluoride uptake may extend the scope of [¹⁸F]NaF PET/CT imaging in FOP from early detection of areas of increased heterotopic bone formation to monitoring disease activity and response to therapeutic interventions. Simplified scanning protocols, without the need for arterial or venous sampling could make it feasible to use [¹⁸F]NaF PET/CT scanning for routine assessment of FOP patients. In a cross-sectional analysis, both SUV_mean and TBR_mean showed a strong correlation with NLR-derived K_i, both before and during treatment with garetosmab. When evaluating the same lesions over time, changes in fluoride uptake as measured by TBR_mean best reflected the change in fluoride uptake as measured by NLR-derived Ki.

To the best of our knowledge, this is the first study performing dynamic [¹⁸F]NaF scans in patients with FOP. Due to the nature of the disease, there are a few considerations that need to be mentioned. Obtaining an image-derived input function through continuous arterial sampling, the preferred option for dynamic scanning, has an unacceptable high risk of causing heterotopic ossification in FOP. We therefore opted for venous sampling, but were restricted to 1 ml per sample, to do as little harm as possible. In 1 ml of whole blood it is possible to accurately measure the [¹⁸F]NaF concentration in MBq/ml, though for conversion to plasma levels a fixed value of 1.21 was assumed based on an in-house unpublished data set for the same tracer from different earlier studies, though other studies have found similar plasma-to whole blood ratios (19). Secondly, most PET/CT scanners currently in use have axial fields of view ranging from 15 to 30 cm, limiting the body area that can be scanned at any given time. By moving the scanner bed during the scan, tracer uptake throughout the body can be visualized, providing a so-called (static) whole body PET/CT scan. In dynamic PET/CT scans, the position of the scanner bed is fixed, as tracer uptake needs to be monitored continuously over the same position. As a consequence, the field of view available for assessment of lesions in dynamic PET/CT scans is limited. To investigate the rest of the body, often a whole static sweep is performed after the dynamic sequence, making it a necessity to assess the performance of simplified parameters such as SUV and TBR. Fortunately, new total body PET/CT scanners have been developed, expanding the field of view to one to two meters. This is of particular interest for conditions where lesions are distributed throughout the body, such as in FOP, and this generation of scanners should therefore be considered for use in future studies when possible.

Despite the limited field of view in our study, we were still able to identify multiple active PET lesions and heterotopic bone in all participants, making it possible to evaluate the correlation between [¹⁸F]NaF kinetics and various semi-quantitative parameters. FOP is a heterogeneous disease spread throughout the musculature of the body and which progresses at an unpredictable pace (8). A piece of heterotopic bone may be growing in one location, whilst remaining stable elsewhere. In the present study, various volumes of interest were defined, i.e., CT-based VOIs, in which heterotopic bone was delineated, and PET-based VOIs in which active lesions were defined based on [¹⁸F]NaF uptake. From a clinical perspective, it makes more sense to follow the PET-active lesions rather than existing stable heterotopic bone with regard to disease activity. These are the active areas that are not bone yet, but may rapidly be so. Moreover, the changes in TBR_mean and SUV_mean better reflected the changes in the net rate of [¹⁸F]NaF uptake, in these PET-active lesions. We recommend future studies to focus on measuring uptake in PET active regions and monitor them over time by measuring TBR. Following (systemic) therapy, the correlation between semi-quantitative measures and true [¹⁸F]NaF kinetics should be re-evaluated, as therapy may affect the underlying biology (e.g., perfusion) in which case simple baseline and follow-up uptake measurements may not be comparable with each other.

The better performance of TBR is likely due to variation in the blood background activity, which is accounted for in the TBR calculation, but not in the SUV calculation. TBR is a semi-quantitative parameter frequently used in PET studies of vascular inflammation and atherosclerotic plaques. Blood activity of the tracer can vary between patients due to biological factors affecting its distribution, uptake and excretion (22). Uptake of an irreversible tracer is directly proportional to perfusion, so both SUV and TBR will be affected by changes in perfusion. Locally, vascularization varies throughout the endochondral ossification process. Initially, a hypoxic poorly vascularized area is present during the cartilaginous phase. This hypoxia, in turn, promotes neovascularization of the cartilage as it transitions to the ossifying phase (6, 23, 24). The normalization of [¹⁸F]NaF uptake to blood activity, being part of TBR, corrects for any variation in [¹⁸F]NaF blood supply towards the target tissue.

In a few cases, [¹⁸F]NaF PET/CT has already been used to evaluate ossifying activity in FOP. Elevated SUV was reported in a FOP patient after jaw surgery, preceding subsequent heterotopic ossification (14). Later, a series of 13 [¹⁸F]NaF PET/CT scans in 5 different patients were examined, and areas with increased ¹⁸F-fluoride uptake were identified, which subsequently corresponded with radiological progression of HO on a later scan (7). Areas with increased fluoride uptake, measured by SUV, were found to be associated with an increase in heterotopic ossification as measured by CT. A SUV_peak higher than 8.4 was found to be indicative of new HO on a later scan, though the increment of the SUV did not appear to correlate with the amount of bone volume formed. The higher correlation of TBR with full kinetic analysis in the present study, suggests that TBR may be better suited as a semi-quantitative parameter for predicting heterotopic bone formation. Ideally, longitudinal studies with multiple [¹⁸F]NaF PET/CT scans need to be performed to determine the optimum cut-off of TBR with regard to sensitivity and specificity for predicting heterotopic ossification in FOP. Most trials in FOP have used the change in total HO volume, as assessed by low-dose whole body CT, as the primary outcome measure for reflecting disease activity and thus drug effectiveness. In addition to the measured HO volume on CT, PET active lesions can be followed over time, visualizing and quantifying disease activity before they are seen on CT images. Other outcome measures reflecting disease activity in FOP include the number of new HO lesions as assessed by CT, number of body regions with HO, flare-up incidence, quality of life questionnaires and clinical measurements, such as CAJIS and pulmonary function tests (NCT03312634, NCT02190747, NCT03188666, NCT04307953). Two trials have also included the number of active PET lesions and lesion activity as outcome measures (NCT03188666, NCT04307953).

Our present study supports TBR as the most suitable semi-quantitative parameter to measure change in ¹⁸F-fluoride uptake in patients with FOP, given the high correlation with NLR-derived K_i and the short scanning time required. In addition, changes in uptake are best evaluated in small PET active regions rather than over larger areas of heterotopic bone. Separate studies with dynamic scanning should be performed if there is reason to believe that ¹⁸F-fluoride kinetics may be altered by the therapy itself, rather than as a result of a change in bone metabolism reflecting disease activity.