Characteristics of HH Patients and Controls

Subject characteristics of all 24 HH patients are presented in Table 1, for 1:1 matched controls in Tables 3 and 4. HH patients were mostly male (about 67%), with a mean age of 54.7 ± 10.8 years and had been diagnosed with HH approximately 10.2 ± 9.2 years ago. 19 out of the 24 HH patients had received regular treatment by phlebotomies, 2 patients had been without treatment and for three HH patients, information on HH-related treatment was not available. On average, HH patients had been treated with about 6 phlebotomies annually and had received approximately 60.6 ± 66.0 phlebotomies since HH diagnosis. With respect to arthralgias, we found that about 71% of all HH patients reported joint pain in at least one hand at the time of the study visit. When looking at sex differences, we noticed that HH men had received numerically more phlebotomies than HH women since diagnosis (p = 0.085). However, all other anthropometrics were comparable between sexes including the proportion of HH individuals with hand arthralgias per group and the time since HH diagnosis.

Table 1 Patient characteristics given for all 24 study participants with HH and stratified by genderSegmentation Performance and Algorithm Completion Rates of Joint Space Analysis in HH Patients

From the 24 HH patients, a total of 72 MCP joints (24 MCP 2, 24 MCP 3, 24 MCP 4 joints) were available to be processed via the HR-pQCT-based UCSF JSW algorithm [26]. Representative axial cross-sectional HR-pQCT images of a HH patient are shown in Fig. 2 and representative 3D visualizations of the joint space segmentation and morphometric results are presented in Fig. 3. Due to poor image quality related to motion, two out of 72 MCP joints were excluded from the analysis, one MCP joint was excluded due to ankylosis of the joint space. For the remaining 69 MCP joints, the joint spaces were successfully segmented automatically for 55 cases at the first attempt (79.7% overall; 75% of MCP 2 joints, 81.8% of MCP 3 joints, 82.6% of MCP 4 joints). Additional semi-manual segmentation intervention to separate the individual bones was required for 14 cases (20.3%). This correction step involved a short (< 5 min) semi-automatic coarse contour of the metacarpal head by the operator to differentiate it from the proximal phalangeal base.

Fig. 2

Axial and coronal HR-pQCT images showing metacarpal heads 2–4 (axial view) and MCP joints 2–4 (coronal view) of a patient with HH. Note the distinct features of hemochromatosis arthropathy: characteristic hook- or beak-like osteophytes at the metacarpal heads are seen (white arrows) along with small cysts. In addition, areas of sparse to absent trabecular bone structure are noted in the center of the metacarpal heads 2 and 3 (dashed white arrow) indicative of a HH-related impairment of bone microarchitecture

Fig. 3

a Initial 3D rendering of the segmentation result of the MCP 4 joint of a 72-years-old male, diagnosed with HH 31 years ago. The metacarpal head is depicted in light blue, while the base of the proximal phalanx is shown in dark blue. The local joint space width was mapped in the joint space in pseudo-color. Blue and green colors reflect a narrow joint space width, while red colors code for a broader joint space width. Note the very narrow and irregular joint space depicted in light green. b Final 3D surface reconstruction of the MCP 3 joint of a 55-years-old male (BMI 21.9 kg/m2) diagnosed with HH 8 years ago. The local joint space width was mapped in the joint space in pseudo-color. Blue and green colors reflect a narrow joint space width, while red colors signify a broader joint space width

When looking at the minimum joint space width (JSW.MIN) needed, above which the JSW software would be able to correctly segment the MCP joints at a success rate of 100% at the first attempt, we found that for all MCP joints, the minimum mean JSW value was 0.082 mm (i.e., one voxel wide). When we evaluated the segmentation performance of the MCP JSW quantification software stratified by sex, we observed that in HH men and HH women similar proportions of MCP joints were successfully automatically segmented by the software (87.5% of all 24 female MCP joints vs. 75.6% of all 45 male MCP joints, p = 0.245). Only 12.5% of MCP joints from female HH patients and 24.4% from male HH patients required additional semi-manual correction (p = 0.245).

Joint Space Parameters Measured by High-Resolution Peripheral Quantitative Computed Tomography (HR-pQCT) in HH Men and Women

With respect to MCP joint morphology, we observed that the joint space volume (JSV) was significantly larger in HH men compared to HH women in all three MCP joints (+ 25.4% at MCP 2, + 34.9% at MCP 3, + 41.8% at MCP 4, 0.001 < p < 0.043). Additionally, mean JSW at the MCP 4 was significantly larger (+ 13%, p = 0.043), and the maximal JSW was significantly higher (up to + 14.7%) in the MCP 3 and 4 (p < 0.011) in HH men than HH women. Also, a strong trend (+ 9.3% JSW.MAX, p = 0.062) toward a larger maximal JSW was observed in the MCP 2 joint in HH men compared to HH women. All other MCP structural and morphological parameters were comparable between HH men and HH women.

Correlations of MCP Joint Morphological Parameters and Clinical Measures in HH patients

With respect to clinical correlations (Table 2), we found that in the overall cohort, time since HH diagnosis was significantly correlated with MCP 4 JSW asymmetry (JSW.AS: ρ = 0.463, p = 0.040) and MCP 4 JSW heterogeneity (JSW.SD: ρ = 0.499, p = 0.025). In addition, the total number of phlebotomies since diagnosis was moderately correlated with the JSW.SD at all MCP sites, reaching statistical significance at MCP 3 and 4 (0.492 < ρ < 0.535, p < 0.045) and approaching significance at MCP 2 (ρ = 0.432, p = 0.073). At MCP 2, there were also statistical trends for correlation between JSW asymmetry and the number of phlebotomies since HH diagnosis (ρ = 0.460, p = 0.055). Estimated therapy intensity as indicated by the yearly rate of phlebotomies was for the overall cohort not significantly correlated with any of the joint structural parameters (data not shown).

Table 2 Spearman´s rho correlations showing the associations between HR-pQCT-derived joint space measures obtained from MCP joints 2, 3, and 4 with clinical parameters in HH men and HH women

In HH men, time since HH diagnosis showed a trend for moderate correlation with JSW heterogeneity at MCP 2 and JSW asymmetry at MCP 3 (0.507 < ρ < 0.558, p < 0.064). With respect to the total number of phlebotomies since HH diagnosis, significant correlations were seen in HH men with joint space volume and minimal JSW at MCP 2 (0.668 < ρ < 0.683, 0.020 < p < 0.025) and with JSW asymmetry and JSW heterogeneity at MCP 3 (0.669 < ρ < 0.703, 0.035 < p < 0.049).

In HH women, no significant correlation between time since HH diagnosis or the number of phlebotomies and MCP joint morphological parameters was found. Estimated therapy intensity as defined by the annual rate of phlebotomies did not significantly correlate with the HR-pQCT-derived MCP joint space parameters (data not shown).

Joint Space Morphology Parameters in HH Patients with and Without Joint Pain

In an exploratory subanalysis, we investigated the differences in MCP joint morphology between HH patients with and without hand arthralgia (see Supplementary Table 1). We found that HH patients with hand arthralgia had a higher mean age (56.1 ± 10.3 years vs. 45.0 ± 15.4 years) and a longer HH duration (11.9 ± 10.2 vs. 5.0 ± 2.0 years) compared to HH patients without pain; however, these differences were not large enough to translate into statistical significances. HH patients with joint pain also showed lower joint space volume (JSV) at all three MCP joints, reaching statistical significance at MCP 2 and MCP 4 (MCP 2: p = 0.009; MCP 4: p = 0.048). Additionally, maximal joint space width (JSW.MAX) was significantly smaller at MCP 2 (p = 0.040) in HH patients with hand arthralgia compared to HH patients without arthralgia.

Differences in MCP Joint Space Morphology Between HH Patients and Controls

In order to better understand the differences in MCP joint morphology between HH patients and healthy controls, we performed additional exploratory subanalyses for which HH patients were matched 1:1 by sex and age to their respective healthy controls. Results of these subanalyses are shown in Tables 3 and 4. HH women exhibited at the MCP 3 joint a significant, 1.8 times larger JSW asymmetry and an about 1/3 larger JSW heterogeneity relative to age-matched healthy control women (MCP 3 JSW.AS: + 180% p = 0.025; JSW.SD: + 37.5%, p = 0.026). Minimum JSW was significantly smaller in HH women compared to controls (MCP 3: JSW.MIN: − 41.8%, p = 0.022) with mean JSW showing also a statistical trend toward being smaller in HH women relative to controls (MCP 3: JSW: − 13.9%, p = 0.086). Similar to the MCP 3 joint, we observed at the MCP 2 joint numerically larger JSW asymmetry and larger JSW heterogeneity and numerically smaller minimal, mean, and maximal JSW in HH women. However, these results did not reach statistical significance with the exception of mean JSW, where a statistical trend toward a smaller mean JSW was noted in HH women (MCP 2: JSW: − 13.3%, p = 0.087).

Table 3 Results from the exploratory subanalysis comparing MCP joint space morphology between HH women and age-matched healthy control womenTable 4 Results from the exploratory subanalysis comparing MCP joint space morphology between HH men and age-matched healthy control men

For HH men versus controls, a similar, but more pronounced and uniform pattern of MCP joint space differences was noted consistently throughout MCP 2 and 3 joints. At both MCP 2 and 3 joints, JSW asymmetry was significantly and around 3 × times larger in HH men relative to healthy men (MCP 2: JSW.AS: + 323%, p = 0.043; MCP 3: JSW.AS: + 359%, p = 0.020), while minimum JSW was at both sites around 2 × smaller (MCP 2: JSW.MIN:− 225%, p = 0.043; MCP 3: JSW.AS: − 216%, p = 0.043). All other joint space parameters did not differ between HH men and their age-matched controls.