Interpretation of plain radiographs document the amount (area) of HO on both the AP and lateral radiograph views. The area for each view was a sum of all the new HO at the flare-up location (and thus if there are multiple HO lesions, the area of each lesion was determined and then the total across all lesions were summed to obtain a total new HO). This total new HO sum was used in the analysis of the area of new HO. Results from Primary Read reviews are presented.

A responder was defined as a subject with no or minimal new HO at the flare-up site versus baseline as assessed by plain radiographs at Week 12. Minimal new HO is defined as new HO with an HO score <=3 in both the AP and lateral projections (or if one view is non-interpretable or non-evaluable, then the remaining evaluable view is used). The HO score ranges from 0 to 6 where, 0 = no HO and 6 = single contiguous HO with longest dimension >2 diameters of the reference normotopic bone in any projection. The highest HO score from the 2 projections was used. Results from the Primary Read reviews are presented. The Primary Read process included a double-read radiology review paradigm with consensus adjudication. Radiography and CT scans were examined independently by scan type, flare-up region, and imaging time point in order to determine whether radiography would be sufficient to measure new HO formation.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Bone specific alkaline phosphatase was analysed as a bone and cartilage biomarker.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. C-reactive protein was analysed as a inflammation biomarker.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. C-terminal telopeptide was analysed as a bone and cartilage biomarker.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Procollagen type 1 N-terminal propeptide was analysed as a bone and cartilage biomarker.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Procollagen type 1 C-terminal propeptide was analysed as a bone and cartilage biomarker.

Low dose CT scan were used as a secondary imaging assessment of HO, and was performed at the same time points as plain radiographs. Interpretation of the CT scan documented the amount (volume) and grade of HO. The independent reviewer scored HO lesions according to the following scale for HO on CT. Grade 1 = fluid attenuation without evidence of calcification at CT, Grade 2 = calcification of soft tissues without evidence of bone formation, Grade 3 = immature bone formation, and Grade 4 = mature bone with cortical differentiation. Volume of new HO was determined according to the following steps: (1) calculate volume of new HO compared to baseline for each reviewer/HO ID, (2) sum the volume of new HO across HO IDs for each reviewer, and (3) average the volume of new HO across reviewers. Results from Primary Read reviews are presented.

The MRI was utilized to evaluate the presence of soft tissue swelling/edema (and volume of the swelling/edema) and presence of cartilage formation (yes or no). For subjects who could not have an MRI, US was used to assess edema severity for the sub-set of subjects enrolled after this opinion was introduced in a protocol amendment. Imaging film from MRI was assessed by two independent readers. When there was sufficient agreement between the independent readers on volume, both of the independent readings were used for analysis with the volume measurements averaged. When there was insufficient agreement between the independent readers, an adjudication reading was provided and used for analysis. The US was used for soft tissue swelling/edema but not cartilage formation. Percentage calculated as % = 100 x n/N' where N' is the number of subjects with interpretable outcomes. Results from Primary Read reviews are presented.

Active range of motion, expressed as the percent of normal arc of motion, measurements at the primary joint associated with the flare-up and adjoining joints was assessed by goniometer.

Flare-up movement outcomes were independently assessed by both the subject (or parent of a subject under 8 years of age) and the Investigator at Weeks 6 and 12 by completing the global assessment of movement. The subject/parent completed the global assessment first. Prior to reviewing the subject's assessment, the Investigator completed his/her own assessment of the flare-up outcome. Subjects were assessed how the flare-up affected their movement on a scale ranging 1 to 5 where, 1 = severely worse movement and 5 = better movement compared with study Day 1 (day of first dose of study drug). Investigators were assessed how the flare-up affected the subject's movement on a scale ranging 1 to 5 where, 1 = severely worse movement and 5 = better movement compared with baseline (day of screening physical examination).

The pain and swelling associated with flare-ups was evaluated using 2 separate numeric rating scales, one for pain and one for swelling. The pain scale ranges from 0 to 10 where, 0 = no pain and 10 = worst pain ever experienced. The swelling scale ranges from 0 to 10 where, 0 = no swelling and 10 = worst swelling ever experienced. The Faces Pain Scale - Revised (FPS-R) was used for children less than 8 years old. The FPS-R ranges from 0 to 10 where, 0 = no pain and 10 = very much pain in two-point increments.

Subjects were given a list of FOP assistive devices and adaptations and asked to select those they use for daily living. The FOP assistive devices and adaptations included mobility aids, care attendants, eating tools, personal care tools/aids, bathroom aids and devices, bedroom aids and devices, home adaptations, work environment adaptations, technology adaptations, sports and recreation adaptations, school, and medical therapies for daily living.

The duration of active symptomatic flare-up was defined as the number of days the subject reported the presence of symptoms in the diary ('Is your flare-up ongoing today?') from Day 1 to study completion at Day 84. The mean number of days of active, symptomatic flare-up is presented for subjects with evaluable diary data.

The FOP-PFQ consists of 28 questions rated on scales from 1 to 5, with lower scores denoting more difficulty. The adult form of the FOP-PFQ was administered to subjects 15 years of age and older. There are two Pediatric FOP-PFQ (FOP-PFQ-P) forms: a self-completed form for 8 to 14 year-olds and a parent proxy-completed form for 5 to 14 year-olds. For subjects between 8 to 14 years of age, both the self-completed (for 8 to 14 year-olds) and the proxy-completed (for 5 to 14 year olds) forms of the FOP-PFQ-P were administered. However, only the proxy-completed form was used for analysis. Percentage of worst scores ranges from 0% to 100% with 0% = best possible function and 100% = worst possible function. Change from baseline for each time point is presented.

The PROMIS Global Health contains 10 questions which are rated on scales from 1 to 5 or 0 to 10. Global physical health scores were calculated as the sum of scores from parameters 3, 6, 7, and 8 and ranges from 4 to 20 where, 4 = worse health and 20 = better health. Global mental health scores were calculated as the sum of scores from parameters 2, 4, 5, and 10 and ranges from 4 to 20 where, 4 = worse health and 20 = better health. For paediatric subjects, the PROMIS was administered as per the adult version. However, there is a single total score for the paediatric PROMIS (as opposed to global physical and global mental health scores as are in the adult version). The total score were converted to a T-score. A T-score of 50 is normal and increments of 10 are +/- 1 standard deviation away from the norm. A T-score <50 indicates worse health, while a T-score >50 indicates better health. The higher values (positive changes) indicate better health.

The Cmax of palovarotene was determined.

The Cmin of palovarotene was determined.

The Tmax obtained by inspection of palovarotene was determined.

The t1/2 was calculated as ln(2)/ λz. The number of data points included in the regression was determined by visual inspection, but a minimum of three data points in the terminal phase, excluding Cmax, was required to estimate λz.

The AUC(0-24hr) was calculated using linear trapezoid rule.

The CL/F was defined as dose/AUC0-24hr.

Subjects received placebo (matching with palovarotene) for 42 days during flare-ups. Subjects were followed for an additional 42 days without treatment. Subjects in Cohort 1 and Cohort 2 contributed to this arm.

Total of all reporting groups

Low dose CT scan was used as a secondary imaging assessment of HO and was performed at the same time points as plain radiographs. The percentage of subjects with new HO (regardless of the amount of new HO) at the flare-up site as assessed by CT scan and/or plain radiographs at Weeks 6 and 12 were analysed. The results are from Global Read reviews. The holistic Global Read process allowed concurrent review of all modalities across all time points, and provided access to selected clinical data at the time of review.

The PP population included all subjects who were eligible for the FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site.

Interpretation of plain radiographs document the amount (area) of HO on both the AP and lateral radiograph views. The area for each view was a sum of all the new HO at the flare-up location (and thus if there are multiple HO lesions, the area of each lesion was determined and then the total across all lesions were summed to obtain a total new HO). This total new HO sum was used in the analysis of the area of new HO. Results from Primary Read reviews are presented.

The PP population included all subjects who were eligible for the FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site.

A responder was defined as a subject with no or minimal new HO at the flare-up site versus baseline as assessed by plain radiographs at Week 12. Minimal new HO is defined as new HO with an HO score <=3 in both the AP and lateral projections (or if one view is non-interpretable or non-evaluable, then the remaining evaluable view is used). The HO score ranges from 0 to 6 where, 0 = no HO and 6 = single contiguous HO with longest dimension >2 diameters of the reference normotopic bone in any projection. The highest HO score from the 2 projections was used. Results from the Primary Read reviews are presented. The Primary Read process included a double-read radiology review paradigm with consensus adjudication. Radiography and CT scans were examined independently by scan type, flare-up region, and imaging time point in order to determine whether radiography would be sufficient to measure new HO formation.

The PP population included all subjects who were eligible for FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site. Only subjects with interpretable outcomes were evaluated.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Bone specific alkaline phosphatase was analysed as a bone and cartilage biomarker.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. C-reactive protein was analysed as a inflammation biomarker.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. C-terminal telopeptide was analysed as a bone and cartilage biomarker.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Procollagen type 1 N-terminal propeptide was analysed as a bone and cartilage biomarker.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Blood and urine samples for analysis of cartilage, bone, angiogenesis, and inflammation biomarkers were collected. Procollagen type 1 C-terminal propeptide was analysed as a bone and cartilage biomarker.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Low dose CT scan were used as a secondary imaging assessment of HO, and was performed at the same time points as plain radiographs. Interpretation of the CT scan documented the amount (volume) and grade of HO. The independent reviewer scored HO lesions according to the following scale for HO on CT. Grade 1 = fluid attenuation without evidence of calcification at CT, Grade 2 = calcification of soft tissues without evidence of bone formation, Grade 3 = immature bone formation, and Grade 4 = mature bone with cortical differentiation. Volume of new HO was determined according to the following steps: (1) calculate volume of new HO compared to baseline for each reviewer/HO ID, (2) sum the volume of new HO across HO IDs for each reviewer, and (3) average the volume of new HO across reviewers. Results from Primary Read reviews are presented.

The PP population included all subjects who were eligible for the FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site.

The MRI was utilized to evaluate the presence of soft tissue swelling/edema (and volume of the swelling/edema) and presence of cartilage formation (yes or no). For subjects who could not have an MRI, US was used to assess edema severity for the sub-set of subjects enrolled after this opinion was introduced in a protocol amendment. Imaging film from MRI was assessed by two independent readers. When there was sufficient agreement between the independent readers on volume, both of the independent readings were used for analysis with the volume measurements averaged. When there was insufficient agreement between the independent readers, an adjudication reading was provided and used for analysis. The US was used for soft tissue swelling/edema but not cartilage formation. Percentage calculated as % = 100 x n/N' where N' is the number of subjects with interpretable outcomes. Results from Primary Read reviews are presented.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Active range of motion, expressed as the percent of normal arc of motion, measurements at the primary joint associated with the flare-up and adjoining joints was assessed by goniometer.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

Flare-up movement outcomes were independently assessed by both the subject (or parent of a subject under 8 years of age) and the Investigator at Weeks 6 and 12 by completing the global assessment of movement. The subject/parent completed the global assessment first. Prior to reviewing the subject's assessment, the Investigator completed his/her own assessment of the flare-up outcome. Subjects were assessed how the flare-up affected their movement on a scale ranging 1 to 5 where, 1 = severely worse movement and 5 = better movement compared with study Day 1 (day of first dose of study drug). Investigators were assessed how the flare-up affected the subject's movement on a scale ranging 1 to 5 where, 1 = severely worse movement and 5 = better movement compared with baseline (day of screening physical examination).

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site. Only subjects with non-missing values were presented.

The pain and swelling associated with flare-ups was evaluated using 2 separate numeric rating scales, one for pain and one for swelling. The pain scale ranges from 0 to 10 where, 0 = no pain and 10 = worst pain ever experienced. The swelling scale ranges from 0 to 10 where, 0 = no swelling and 10 = worst swelling ever experienced. The Faces Pain Scale - Revised (FPS-R) was used for children less than 8 years old. The FPS-R ranges from 0 to 10 where, 0 = no pain and 10 = very much pain in two-point increments.

The PP population included all subjects who were eligible for the FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site.

Subjects were given a list of FOP assistive devices and adaptations and asked to select those they use for daily living. The FOP assistive devices and adaptations included mobility aids, care attendants, eating tools, personal care tools/aids, bathroom aids and devices, bedroom aids and devices, home adaptations, work environment adaptations, technology adaptations, sports and recreation adaptations, school, and medical therapies for daily living.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

The duration of active symptomatic flare-up was defined as the number of days the subject reported the presence of symptoms in the diary ('Is your flare-up ongoing today?') from Day 1 to study completion at Day 84. The mean number of days of active, symptomatic flare-up is presented for subjects with evaluable diary data.

The PP population included all subjects who were eligible for the FAS population, completed Week 6 study visit with no major protocol deviations with at least 80% compliance with study drug, and had an evaluable radiograph or CT at Week 6 sufficient to allow determination of HO at flare-up site.

The FOP-PFQ consists of 28 questions rated on scales from 1 to 5, with lower scores denoting more difficulty. The adult form of the FOP-PFQ was administered to subjects 15 years of age and older. There are two Pediatric FOP-PFQ (FOP-PFQ-P) forms: a self-completed form for 8 to 14 year-olds and a parent proxy-completed form for 5 to 14 year-olds. For subjects between 8 to 14 years of age, both the self-completed (for 8 to 14 year-olds) and the proxy-completed (for 5 to 14 year olds) forms of the FOP-PFQ-P were administered. However, only the proxy-completed form was used for analysis. Percentage of worst scores ranges from 0% to 100% with 0% = best possible function and 100% = worst possible function. Change from baseline for each time point is presented.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

The PROMIS Global Health contains 10 questions which are rated on scales from 1 to 5 or 0 to 10. Global physical health scores were calculated as the sum of scores from parameters 3, 6, 7, and 8 and ranges from 4 to 20 where, 4 = worse health and 20 = better health. Global mental health scores were calculated as the sum of scores from parameters 2, 4, 5, and 10 and ranges from 4 to 20 where, 4 = worse health and 20 = better health. For paediatric subjects, the PROMIS was administered as per the adult version. However, there is a single total score for the paediatric PROMIS (as opposed to global physical and global mental health scores as are in the adult version). The total score were converted to a T-score. A T-score of 50 is normal and increments of 10 are +/- 1 standard deviation away from the norm. A T-score <50 indicates worse health, while a T-score >50 indicates better health. The higher values (positive changes) indicate better health.

The FAS population included all subjects who received at least 1 dose of study drug and had at least 1 evaluable post-baseline radiograph or CT sufficient to allow determination of HO at the flare-up site.

The Cmax of palovarotene was determined.

The Pharmacokinetic (PK) population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

The Cmin of palovarotene was determined.

The PK population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

The Tmax obtained by inspection of palovarotene was determined.

The PK population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

The t1/2 was calculated as ln(2)/ λz. The number of data points included in the regression was determined by visual inspection, but a minimum of three data points in the terminal phase, excluding Cmax, was required to estimate λz.

The PK population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

The AUC(0-24hr) was calculated using linear trapezoid rule.

The PK population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

The CL/F was defined as dose/AUC0-24hr.

The PK population included all subjects who had sufficient blood samples collected for valid estimation of PK parameters.

Subjects received palovarotene 5 mg orally for 14 days followed by 2.5 mg orally for 28 days during flare-ups (5/2.5 mg regimen). The subjects were followed for an additional 42 days without treatment. Only subjects in Cohort 2 contributed to this arm.

Subjects received placebo (matching with palovarotene) for 42 days during flare-ups. Subjects were followed for an additional 42 days without treatment. Subjects in Cohort 1 and Cohort 2 contributed to this arm.

Any flare-ups reported during the 12-week duration of the study, which were not the flare-up that qualified the subject for the study, were recorded as adverse events.