ACHILLES is a two-treatment-arm, randomised, parallel-group, double-blind, placebo-controlled study in patients with PsA and axial SpA (Additional file 1). Patients were randomised in a ratio of 1:1 to receive either secukinumab or matching placebo at baseline, Weeks 1, 2, 3 and 4 and once every 4 weeks thereafter. Patients on placebo switched to active secukinumab treatment starting at Week 24.
A total of 204 patients (age ≥ 18 years; 102 in the secukinumab 150 or 300 mg group and 102 in the placebo group) with active SpA (peripheral or axial) were included in the ACHILLES trial if they presented with clinical and MRI-positive heel enthesitis: clinical heel enthesitis defined as swelling and tenderness at the insertional site of the Achilles tendon into the calcaneus (binary pain assessment present/absent), and MRI-positive heel enthesitis defined as tendinitis with/without bursitis and/or bone marrow oedema (BME) with/without concomitant erosions in the insertional area of the Achilles tendon and/or the plantar aponeurosis. MRI-positive heel enthesitis was interpreted by either the local radiologist or rheumatologist at the individual study site.
MRI of the affected foot/ankle was performed by imaging technicians on standard MRI systems with magnetic field strengths of at least 1.5 Tesla and with the technical capability of performing foot MRI examinations. The MRI study protocol was standardized for all participating centres in an imaging manual and consisted of two mandatory sequences (T1-weighted turbo spin-echo/fast spin-echo in sagittal and transversal orientation and short inversion time inversion-recovery [STIR] in sagittal and transversal orientation) (Additional file 2).
MRIs were performed at three time points: screening, Week 24 and Week 52. No preparative drugs, contrast agents or radionuclide agents were used during the procedure.
Assessments of heel MRI characteristics
Heel MRI was assessed both qualitatively and quantitatively. The qualitative assessment of heel MRI characteristics (absent/present) was performed as follows (Table 1): (i) tendinitis in the Achilles tendon and/or plantar aponeurosis, (ii) tendinitis in the Achilles tendon and/or plantar aponeurosis: insertional area, (iii) bursitis in the area of the Achilles tendon and/or plantar aponeurosis, (iv) BME in the area of the Achilles tendon and/or plantar aponeurosis, (v) periarticular inflammation of the ankle joint and (vi) bone erosion.
The quantitative assessment of heel MRI characteristics (based on PsAMRIS) is as follows (Table 2): (i) bone oedema; proportion of bone oedema was compared to the volume of the affected bone: 0 = absent, 1 = 1–33% of bone oedematous, 2 = 34–66% of bone oedematous, 3 = ≥ 67% of bone oedematous and (ii) bone erosion; proportion of bone erosion was compared to the volume of the affected bone: 0 = absent, 1 = 1–10% of bone eroded, 2 = 11–20% of bone eroded, …, 10 = 91–100% of bone eroded.
The mean scores of heel MRI characteristics such as bone oedema quantification and bone erosion quantification will be calculated. The mean score of bone oedema quantification will be calculated based on the individual scores of each subject (0 = absent, 1 = 1–33%, 2 = 34–66%, 3 = ≥ 67%). Similarly, the mean score of bone erosion quantification will be calculated based on the individual scores of each subject (0 = absent, 1 = 1–10%, 2 = 11–20%, …, 10 = 91–100%).
The mean composite score based on PsAMRIS will be calculated based on the individual scores of each subject for periarticular inflammation, bone oedema and bone erosion as follows: periarticular inflammation (0 = absent, 1 = present), bone oedema (0 = absent, 1 = 1–33%, 2 = 34–66%, 3 = ≥ 67%) and bone erosion (0 = absent, 1 = 1–10%, 2 = 11–20%, …, 10 = 91–100%). Similarly, the mean composite score based on PsAMRIS, extended for bursitis and tendinitis, will be calculated based on the individual scores of each subject for the following parameters: periarticular inflammation (0 = absent, 1 = present), bone oedema (0 = absent, 1 = 1–33%, 2 = 34–66%, 3 = ≥ 67%), bone erosion (0 = absent, 1 = 1–10%, 2 = 11–20%, …, 10 = 91–100%), bursitis (0 = absent, 1 = present) and tendinitis (0 = absent, 1 = present).
Active inflammation was defined as ‘yes’ if at least one of the following parameters was present: tendinitis (Achilles tendon/plantar aponeurosis insertion), bursitis (Achilles tendon/plantar aponeurosis), bone oedema (Achilles tendon/plantar aponeurosis), periarticular inflammation and bone oedema (PsAMRIS). Active inflammation will be determined automatically by the reading software (Table 3) after completion of the manual read process. In case one or more read parameters cannot be determined for a data set (i.e. one image), the readers have the option to mark the parameters as ‘N/A’ (not applicable) and comment on the reasons for their decision.
Adaptations to the heel
Since PsAMRIS was developed initially for the hands, additional parameters were introduced to allow for a more accurate representation of the heel in terms of quantifying bone oedema and locating bone erosion. Bone oedema with 1–33% of bone oedematous was further specified by oedema length ≤/> 0.5 cm.
Bone erosions were further specified, if categorised as at least one and not more than 10, by assessing the localisation as follows: bone erosion in the area of the Achilles tendon, bone erosion in the area of the plantar aponeurosis and bone erosion in the area ‘Other’.
MRIs of bone oedema and bone erosion of patients with active SpA are presented in Fig. 1; heel-specific adaptations are indicated with an arrow.
Central reading scheme
The central read process was implemented in a consensus-read fashion with two readers, blinded for the identification of study centre, patient, treatment or visit/date of scan. The conditions for a consensus read depended on the reading scores: (i) > 1 deviation in the PsAMRIS scoring of the parameters ‘Quantification of Bone Oedema’ and ‘Quantification of Bone Erosion’ and (ii) discrepancy in any of the qualitative reading parameters. In case of a deviation in scoring for ‘Quantification of Bone Oedema’ and ‘Quantification of Bone Erosion’ of 1, the average value was recorded.
To start the read process, a reading task was assigned to the readers on the reading system. One reading task was defined as the image data of all available MRI time points of one patient. The MRIs of one reading task were reviewed according to the scoring forms for each time point. Each reading task was organised in a randomised independent temporal presentation, i.e., images regarding MRI time points were presented to the readers in a blinded fashion.
The detailed reading scheme is presented in Fig. 2. A brief reading scheme is as follows: (i) First Reads: Each reading task was initially rated by each reader, (ii) Asymmetric Consensus Reads: If no consensus was achieved in the First Reads, the case was randomly assigned to one of the readers. This reader now had the opportunity to change his own scores, while previous scores of both readers were displayed. If no consensus scoring was achieved after this step, the case was assigned to the other reader who then had the opportunity to change his scores, while previous scores of both readers were displayed and (iii) Symmetric Consensus Reads: If no consensus was achieved, the case was assigned to both readers simultaneously, and they have to agree to a consensus score by consultation to finish the read.
The clinical information of individual patients was not available to the readers in the read process, and there was no time limit for the completion of a single read.