The triangular fibrocartilage complex on high-resolution 3 T MRI in healthy adolescents: the thin line between asymptomatic findings and pathology

Healthy girls and boys aged 12–18 years without wrist pain over the last 6 months were included. Participants were not eligible if they had participated in gymnastics or performed wrist-loading sports over two times a week, had fused distal radial physes, were diagnosed with growth disturbance or musculoskeletal diseases, or had a history of wrist fracture, surgery, or infection.

All participants underwent a radiograph followed by MRI of one wrist. Conventional posterior-anterior radiographs were obtained in 90° shoulder abduction, 90° elbow flexion, and neutral forearm position (focus-detector distance 1.30 m). Skeletal age was determined using the validated BoneXpert software (v2.0.1.3; Visiana, Holte, Denmark). Ulnar variance was assessed on the radiographs by one specialized musculoskeletal radiologist (SJ) using the recommended perpendicular method [23]. Negative ulnar variance indicates a relatively shorter ulna. MRIs were obtained with a 3-T MRI scanner (Ingenia, Philips Healthcare, Best, The Netherlands) in a feet-first, supine position with the wrist placed neutral, alongside the body, using a dedicated wrist coil (eight channel, receive-only). The analyzed MRI sequences included turbo spin-echo (TSE) proton-density (PD) weighted sequences in three planes without fat saturation (1500–2000 ms repetition time (TR), 20 ms echo time (TE), 90° flip angle, 2.5–1.5 mm slice thickness, 0.30 × 0.30 mm spatial resolution, 4:06–4:39 min scan time), a TSE PD weighted coronal sequence with fat saturation by spectral attenuated inversion recovery (SPAIR) (2000 ms TR, 30 ms TE, 90° flip angle, 2.5 mm slice thickness, 0.30 × 0.32 mm spatial resolution, 4:12 min scan time), and a TSE T2 weighted axial sequence with fat saturation by SPAIR (3001 ms TR, 60 ms TE, 90° flip angle, 2.5 mm slice thickness, 0.30 × 0.36 mm spatial resolution, 4:58 min scan time).

A standardized scoring form focusing on observation of morphology, homogeneity, and signal intensity and explicitly not on detecting pathology was constructed by a physician experienced in research on musculoskeletal imaging (AP). Due to the lack of a consensus reference standard regarding the individual TFCC components, the items were based on MRI definitions used by Zhan et al. on adults, i.e., triangular fibrocartilage (TFC), dorsal and volar radioulnar ligaments (RUL), meniscus homolog, extensor carpi ulnaris (ECU) tendon sheath, ulnocarpal ligaments (ulnolunate and ulnotriquetral), and ulnar collateral ligament [24]. Images from trial cases that were not included in the present study (i.e., symptomatic gymnasts) were provided as illustrative examples in the scoring form (see supplementary material).

First, scoring items were adjusted, added, and deleted based on clinical relevance and feasibility in several rounds of consensus meetings with a single-center expert group of three musculoskeletal radiologists (MM, FS en SJ with 26, 5 and 1 years of experience, respectively) and one hand surgeon (MO with 15 years of experience). Items on ulnocarpal ligaments and the ulnar collateral ligament were deleted due to lack of clinical relevance and controversy over their existence [25]. The recommended sequences and planes for assessment were determined in consensus and stated per item. The final score form, as provided by the supplementary material, was piloted in several trial cases for calibration prior to scoring.

Then, each member of the expert group individually assessed the blinded MRI scans in a randomized order on a diagnostic PC workstation with a high-resolution monitor using the IMPAX software version 6.6.1.4024 (AGFA HealthCare N.V., Mortsel, Belgium). In order to prevent response fatigue, observers were allowed to assess only three consecutive MRIs followed by a mandatory break. Each observer was blinded for other observers’ scoring results and observers were aware that subjects were asymptomatic. Windowing, zooming, and scrolling through the images was allowed.

Patient characteristics and scoring form observations were entered into Castor Electronic Data Capture (Ciwit BV, Amsterdam, The Netherlands, 2018) and directly imported for analysis in RStudio (RStudio, Inc., Boston). Descriptive statistics and TFC thickness were illustrated as medians with ranges. TFC thickness measurements that were not possible due to disc disruption were excluded from this analysis. An intraclass correlation coefficient (ICC) for inter-observer agreement on TFC thickness was calculated using a two-way random-effects (agreement, single measures) model. For each MRI characteristic, categorical variables scored by all observers for all participants were combined and expressed in frequencies of observations with percentages (n = 92). Inter-observer agreement was determined by the unweighted Fleiss’ kappa with 95% confidence intervals (95% CI). All correlation coefficients were interpreted as poor (≤ 0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), or excellent (0.81–1.00) agreement [26]. Differences were calculated with a Mann-Whitney U test. P values <0.05 were considered statistically significant.

The study cohort consisted of 23 healthy adolescents and their demographic information is depicted in Table 1. One boy was excluded due to insufficient image quality based on motion artifacts.

All results for MRI characteristics of TFCC components are shown in Table 2.

Median TFC thickness was 1.4 mm (range 0.1–2.9). Agreement between observers for measuring TFC thickness was moderate (ICC = 0.60, 95% confidence interval = 0.40–0.77). In the coronal plane, TFC morphology was identified as a slightly radially tilted asymmetrical bowtie in 46 (50%) observations, as a shorter, thicker, and more horizontal structure in 34 (37%) and as a thinner and more stretched structure in only 12 (13%) (Fig. 1). We found no statistical difference in wrist position (p = 0.26) or ulnar variance (p = 0.17) for wrists where a thinner and more stretched TFC configuration was observed. In the sagittal plane, TFC morphology was scored as a symmetrical biconcave disc in 48 (52%) and as an asymmetrical disc being thicker at the dorsal side than at the volar side in 41 (45%) observations. Inter-observer agreement for scoring coronal morphology was fair (ICC = 0.33, 95% CI = 0.20–0.46) and for scoring sagittal morphology was poor (ICC = 0.19, 95% CI = 0.20–0.53).

TFC homogeneity in the coronal plane was scored as a homogeneous hypo-intense signal intensity in 43 (47%) and as a diffuse increased signal intensity not reaching the articular surface in 30 (33%) (Fig. 2). In 14 (15%) observations regarding TFC homogeneity, a vertical linear increased signal intensity with discontinuation of the TFC was identified. Other types of homogeneity that were scored included slight morphological variations on this vertical line of increased signal intensity. Inter-observer agreement for coronal homogeneity was poor (ICC = 0.13, 95% CI = 0.16–0.24).

In the sagittal plane, the dorsal and volar RUL were scored as continuous and indiscernible structures with the TFC in 74 (80%) and 55 (60%) observations, respectively, while the volar RUL signal intensity appeared to be discontinuous with the TFC in 23 (25%) observations (Fig. 3). The inter-observer agreement was fair (ICC = 0.23, 95% CI = 0.11–0.36) for volar RUL and poor (ICC = 0.20, 95% CI = 0.07–0.32) for dorsal RUL sagittal morphology. Axial RUL fiber continuity was scored as continuous in 77 (84%) observations of the dorsal RUL and 82 (89%) of the volar RUL. In 5 (5%) observations, the dorsal RUL showed fiber discontinuation, for the volar RUL fiber discontinuation was seen in 1 (1%) observation. For this item, inter-observer agreement regarding the dorsal and volar RUL was poor (ICC = 0.06, 95% CI = −0.07–0.19 and ICC = −0.04, 95% CI = −0.19–0.12, respectively).

Homogeneity of the proximal and distal lamina on the coronal view was scored as homogeneous hypo-intense in 51 (55%) (Fig. 2a) and 44 (48%) observations, respectively. In 33 (36%) of the proximal and 39 (42%) of the distal lamina, a diffuse lamination with hyper-intense signals was seen. Inter-rater agreement for both proximal and distal lamina coronal homogeneity was poor (ICC = 0.18, 95% CI = 0.04–0.32 and ICC = 0.13, 95% CI = −0.01–0.26, respectively).

Coronal morphology of the ligamentum subcruentum was identified as a hyper-intense signal intensity in between the proximal and distal lamina in 74 (80%) observations (see Fig. 2a). In 18 (20%) observations, this structure was not clearly visible. Agreement for this item was fair (ICC = 0.22, 95% CI = 0.05–0.38).

Meniscus homolog morphology on the coronal view was scored as a diffuse triangular shaped hypo-intensity directly ulnar from the prestyloid recess in 52 (57%) observations (Fig. 1c). In 24 (26%) observations, this triangular-shaped hypo-intensity was clearly demarcated. The meniscus homolog was not discernable and therefore scored as not visible in 16 (17%) observations. Inter-observer agreement was poor (ICC = 0.04, 95% CI = −0.08–0.16).

The results of additional TFCC-related MRI characteristics of the ulnar side of the wrist are also shown in Table 2.

Neutral wrist position based on the axial plane was seen in 42 (46%) observations. The wrist was defined as supinated in 47 (51%) observations and pronated in 3 (3%). The ECU tendon position was scored as completely positioned within its groove in 51 (55%), as partially within its groove in 31 (34%), and as completely dislocated out of its groove in 10 (11%) observations (Fig. 4). The distribution of ECU tendon positions significantly differed between a supinated and a non-supinated wrist position (p = 0.031, Table 3). Inter-rater agreement for wrist and ECU tendon position was, respectively, substantial (ICC = 0.62, 95% CI = 0.53–0.84) and moderate (ICC = 0.59, 95% CI = 0.46–0.72).

The signal intensity surrounding and inside the ECU tendon in the axial plane was assessed on the slices in between the proximal base of the ECU groove and the distal border of the extensor retinaculum. The signal intensity was scored as hypo-intense to intermediate in 48 (52%) and 49 (53%) observations, respectively. In 44 (48%) of the peri-tendinous and 43 (47%) of the intra-tendinous signal intensities, focal or linear increased signal intensity was observed. Peri-tendinous increased signal was located proximal from the styloid process in 34 (77%) observations. Intra-tendinous increased signal was located at the styloid level in 29 (67%). Inter-observer agreement for scoring ECU tendon signal intensity was fair for peri-tendinous (ICC = 0.22, 95% CI = 0.05–0.38) as well as intra-tendinous (ICC = 0.29, 95% CI = 0.12–0.45) signal intensity.

Coronal morphology of the prestyloid recess was scored as a conical shape in 37 (40%) observations, a tubular shape in 17 (18%), and a saccular shape in 10 (11%) (Fig. 5). In 23 (30%) assessments, the prestyloid recess itself was not visible. A fair (ICC = 0.31, 95% CI = 0.21–0.42) inter-observer agreement for this item was found. A small amount of distal radioulnar joint (DRUJ) and pisotriquetral joint (PTJ) effusion was present according to 37 (40%) and 44 (48%) observations, respectively. In 12 (14%) and 26 (28%) observations, a substantial amount of DRUJ and PTJ effusion was identified, respectively. Inter-observer agreement was fair (ICC = 0.30, 95% CI = 0.18–0.43) for DRUJ effusion and moderate (ICC = 0.43, 95% CI = 0.31–0.55) for PTJ effusion. Additional ulnar-sided cysts were identified at the volar side in 8 (9%) and on the dorsal side in 3 (3%) observations, with a fair (ICC = 0.36, 95% CI = 0.23–0.50) inter-observer agreement.

Two observers noticed physeal edema in the distal radius and ulna from one participant and three observers noticed a type two lunate bone with bone marrow edema in one participant. Besides several radial and carpal ganglion cysts, no other relevant findings were observed.