Determination of body composition by dual x‐ray absorptiometry in persons with haemophilia

Abstract

Background

There is limited research on body composition in persons with haemophilia (PwH). The literature describes an increased body fat distribution and decreased lean mass in PwH compared to healthy controls using bioimpedance analysis. Using dual x‐ray absorptiometry (DXA), which is known to be the most accurate method, this investigation aims to postulate reference data for body composition parameters within haemophilia severity phenotypes and age groups.

Methods

Persons underwent whole body DXA screening using Horizon. Body fat percentage, estimated visceral adipose tissue (VAT), appendicular fat and lean mass, and lean and fat mass in relation to body height were assessed. Haemophilia severity and five age groups were distinguished.

Results

Two hundred and one persons with mild (n = 44), moderate (n = 41), or severe (n = 116) haemophilia A/B (median age 40 [28–55; 1.IQ–3.IQ] years) were analysed. The median body fat percentage was 28.7% [25.5%–33.9%] and median estimated VAT was 657 g [403–954 g] with no significant difference between severity phenotypes (p = .474; p = .781). Persons with severe haemophilia had less lean mass compared to moderate and mild haemophilia (p = .013; p = .034). Total and appendicular fat is increased in older PwH (aged ≥40 years) compared to younger PwH (aged ≤29 years; p < .05). Lean mass did not differ between age groups.

Conclusion

This study provides valuable reference data for body composition parameters in PwH. Persons with severe haemophilia show significantly less lean mass compared to persons with moderate or mild haemophilia. Body fat percentage and VAT did not differ between severity phenotypes, but increased with age.

1. INTRODUCTION

Persons with haemophilia (PwH) are known to be affected by arthropathy due to joint bleeding, which is more common in persons with severe haemophilia than in persons with moderate or mild hemophilia. ¹ , ² , ³ Many studies evaluated arthropathy in haemophilia since this is the persons’ main restriction going along with pain, loss of function and decrease of quality of life. ⁴ , ⁵ , ⁶ These symptoms frequently add up to a reduction of physical activity, which in turn has a negative impact on, among others, the cardiometabolic system. ⁷ , ⁸ Due to the reduction of joint function, relieving postures or intramuscular haemorrhage, arthropathy frequently goes along with muscle atrophy. ⁹ Therefore, it is assumed that PwH, especially those affected by severe haemophilia show altered body composition compared to subjects without bleeding disorders. There are only few studies dealing with body composition in PwH using bioimpedance analysis, revealing that PwH show increased fat and decreased leg muscle mass compared to healthy controls. ⁸ , ¹⁰ , ¹¹ , ¹² , ¹³ It has been shown, that the overall prevalence of obesity is increased in haemophilic populations. However, this depends to a certain extent on the country of origin and is not transferable to the general population. ⁷ , ¹⁴ , ¹⁵ Both obesity and reduced muscle mass are linked to increased risk for cardiometabolic diseases or insulin resistance for instance. ¹⁶ The risk profile is further dependent on other contributing body composition parameters as in trunk accentuated (android) fat distribution as well as body fat percentage in relation to height. ¹⁷ To get an insight into these parameters different tools can be used, such as calipometry, bioimpedance, infrared analysis or dual x‐ray absorptiometry (DXA), of which DXA is considered the gold standard regarding body composition analysis. ¹⁸ , ¹⁹ In general, body composition analysis is used for both physiological rationales, for example, to examine elite athletes but also to investigate pathological objectives in different patient cohorts. Among others, it provides estimated values for an individual's body fat percentage, visceral adipose tissue (VAT), or lean mass as surrogate measure of skeletal muscle. ²⁰ Literature provides parameters for both the whole body and different subregions, that is, extremities and trunk, which are yet to be described in PwH. ²¹ , ²² , ²³

This explorative study aims to provide severity‐ and age‐related reference data of body composition parameters for PwH. By using DXA, data of body fat and lean mass for both overall body and subregions are provided, which might enhance patient care as comprehensive understanding of body composition can help tailoring treatment plans, nutritional advice, and physical therapy programs.

2. METHODS

2.1. Study design and participants

This investigation was part of a large prospective single‐centre cohort study examining the relation between haemophilia and osteoporosis (registered at clinicaltrials.gov (ID: NCT04524481)). Data were gathered between 2019 and 2022 at the University Hospital Bonn. Adult Caucasian male persons with either mild, moderate, or severe haemophilia A or B were included. Persons with other bleeding disorders or younger than 18 years were excluded from this investigation. This study was conducted in accordance with the principles of good clinical and ethical practice and was approved by the local ethic committee (339/19). Along with the Declaration of Helsinki, all participants gave written informed consent.

2.2. Data acquisition

Persons underwent a 7‐min whole body DXA screening. This assessment was conducted and analysed using Horizon (Hologic, USA; Apex Software; Auto Whole‐Body V. 13.6.7). Fat and lean mass in relation to body size (kg/m²) as well as body fat percentage (percentage of lipids) were measured for the whole body. the software analysis estimated VAT in grams and evaluated android and gynoid fat distribution. ²⁴ Android fat is considered as trunk accentuated fat, while gynoid fat accumulates around the hips, breasts and thighs. It also calculated the android/gynoid ratio. Additionally, the four extremities are also examined regarding the total amount fat and lean mass, displayed in grams. The BMI was calculated (kg/m²) and classified according to the World Health Organization. ²⁵ Joint status was examined using the Haemophilia Joint Health Score (HJHS, Version 2.1; maximum score 124 points, 20 points × 6 joints, plus 4 points assigned to global gait; one joint can reach 0−20 points; higher values indicating worse joint status). Anamnesis including medical history and pharmacological treatment regime was assessed.

2.3. Statistics

The IBM© Statistical Package for Social Sciences 29 (Armonk, NY, USA) was used for calculations. A significance level of p ≤ .05 (95% confidence interval (95%‐CI)) was established. A descriptive analysis of the body composition data was conducted. The body composition data were analysed by severity and independently by age group (18–29, 30−39, 40−49, 50−59 and 60−80). Data were not normally distributed so that the dependent variables were compared within the three different severities as well as within the five age groups using the Kruskal–Wallis‐test. In case of significant differences, Bonferroni correction was used for post hoc testing to account for multiple comparisons.

3. RESULTS

Two hundred and ten participants were recruited for this study. Due to technical issues or personal reasons, data of 201 persons with either mild (n = 44), moderate (n = 41) or severe (n = 116) haemophilia A or B with a median age of 40[28–55] years were analysed. Persons with severe haemophilia had a significantly worse joint status with a mean HJHS of 21.6 ± 18.6 compared to persons with moderate (9.6 ± 8.8, p = .001) or mild (10.0 ± 10.2, p < .001) haemophilia. The most affected joints were ankles (5.3 ± 5.4) followed by elbows (4.7 ± 4.8) and knees (3.2 ± 4.7). An overview of demographic and haemophilia‐specific data are provided in Table 1.

TABLE 1.

Characteristics of included persons with haemophilia.

Variables	Total (n = 201)	Severe (n = 116)	Moderate (n = 41)	Mild (n = 44)	p‐value
Age (years) M ± SD Median [IQ25–75]	43.2 ± 15.5 40 [28–55]	40.3 ± 14.4 37 [27–52]	43.9 ± 15.1 45 [29–57]	44.7 ± 18.3 45 [28–60]	.298
Weight (kg) M ± SD Median [IQ25–75]	85.3 ± 15.9 83 [76–92]	84.6 ± 17.6 81 [75–90]	87.7 ± 15.5 86 [77–95]	85.0 ± 10.3 83 [78–90]	.304
Height (m) M ± SD Median [IQ25–75]	1.80 ± 0.06 1.80 [1.76–1.85]	1.80 ± 0.06 1.80 [1.76–1.85]	1.81 ± 0.07 1.80 [1.76–1.87]	1.79 ± 0.06 1.80 [1.76–1.84]	.829
Haemophilia A (n[%])	175 (87.1)	104 (89.7)	33 (80.5)	38 (86.4)	n/a
Haemophilia B (n[%])	26 (12.9)	12 (10.3)	8 (19.5)	6 (13.6)	n/a
Treatment regime					n/a
Prophylaxis (n[%])	122 (60.7)	109 (94.0)	12 (29.3)	1 (2.3)
On‐demand (n[%])	68 (33.8)	3 (2.6)	25 (61.0)	40 (93.0)
Missing (n[%])	11 (5.5)	4 (3.4)	4 (9.8)	3 (6.8)
HIV (n[%])	27 (13.4)	22 (19.0)	2 (5.1)	3 (7.0)	n/a
Hepatitis C (n[%])	20 (10.0)	15 (12.9)	4 (10.3)	1 (2.3)	n/a
HJHS (score) M ± SD Median [IQ25–75]	16.7 ± 16.5 10.0 [5.0–27.5]	21.6 ± 18.6 a 18 [5.0–33.0]	9.6 ± 8.8 a 7.0 [4.0–11.0]	10.0 ± 10.2 a 6.0 [4.0–13.0]	.001 b

Abbreviations: BMI, body mass index; HJHS, haemophilia joint health score (2.1); n/a, not applicable.

^a Indicates significant Bonferroni post‐hoc analysis (severe‐moderate: p = .001, severe‐mild: p < .001).

^b Indicates significant difference.

Table 2 provides an overview of BMI and raw data of body composition parameters of the included PwH and further differentiates within the three haemophilia severities. Of 201 PwH, 85 (42.3%) had normal weight (BMI between 18.5 and 24.9) while three persons (1.5%) were considered underweight (BMI < 18.5). Eighty‐four (41.8%) PwH were overweight showing BMI scores between 25.0 and 29.9. Twenty‐two (11.0%) persons were categorized into obesity class I corresponding to a BMI between 30.0 and 34.9. Furthermore, three (1.5%) PwH had a BMI between 35.0 and 39.9, being classified as obesity class II and four (1.9%) PwH were classified as obesity class III having a BMI of >40.0. ²⁶ The mean BMI value did not differ between the severity phenotypes (p = .308).

TABLE 2.

Overall body composition determined with single x‐ray parameters of different severity phenotypes.

Variable	Total (n = 201)	Severe (n = 116)	Moderate (n = 41)	Mild (n = 44)	p‐value
BMI (kg/m2) M ± SD Median [IQ25–75]	26.1 ± 4.5 25.3 [23.4–28.0]	25.8 ± 4.9 25.2 [23.3–27.7]	26.5 ± 4.4 25.8 [23.6–27.9]	26.3 ± 4.4 25.8 [23.9–29.5]	.414
Fat/height2 (kg/m2) M ± SD Median [IQ25–75]	7.9 ± 3.9 7.2 [5.9–9.1]	8.1 ± 4.7 7.0 [5.8–9.3]	7.7 ± 2.9 7.2 [6.0–8.7]	7.5 ± 2.6 7.4 [5.6–9.0]	.987
Lean/height2 (kg/m2) M ± SD Median [IQ25–75]	17.5 ± 4.9 17.0 [15.6–18.3]	17.4 ± 6.2 a 16.7 [15.2–18.2]	17.5 ± 2.4 17.1 [15.8–18.8]	17.6 ± 1.8 a 17.8 [16.2–18.9]	.033
Body fat percentage (%) M ± SD Median [IQ25–75]	29.2 ± 6.2 28.7 [25.5–33.9]	29.7 ± 6.5 29.3 [25.5–34.8]	29.0 ± 5.7 27.7 [25.7–32.5]	28.2 ± 5.7 28.9 [24.2–32.5]	.474
Est. VAT mass (g) M ± SD Median [IQ25–75]	737 ± 431 657 [403–954]	731 ± 442 602 [403–953]	754 ± 384 681 [427–1002]	736 ± 453 646 [382–945]	.781
Android fat (g) M ± SD Median [IQ25–75]	2476 ± 1370 2216 [1533–3106]	2491 ± 1502 2210 [1473–3222]	2500 ± 1243 2404 [1695–2897]	2413 ± 1120 2273 [1707–3074]	.970
Gynoid fat (g) M ± SD Median [IQ25–75]	3851 ± 1539 3661 [2884–4311]	3894 ± 1646 3670 [2831–4359]	3898 ± 1531 3703 [2949–4130]	3694 ± 1245 3548 [2726–4243]	.862
Android/Gynoid ratio M ± SD Median [IQ25–75]	0.6 ± 0.2 0.6 [0.4–0.7]	0.6 ± 0.2 0.6 [0.5–0.7]	0.6 ± 0.2 0.6 [0.5–0.7]	0.7 ± 0.2 0.6 [0.4–0.8]	.621

Note: Data displayed as mean ± SD and median [Interquartile range 25–75].

Kruskal–Wallis Test was used.

Abbreviations: Est., estimated; VAT, Visceral adipose tissue.

^a Indicates significant Bonferroni post‐hoc analysis in lean/height² between persons with severe and mild haemophilia (p = .037).

Overall, PwH showed a median body fat percentage of 28.7% [25.5–33.9]. Regarding the VAT, a median of 657 g [403–954] was found. The analysis revealed no significant difference in overall body composition parameters between the severity phenotypes, except for the lean mass index (lean/height²; p = .033). Bonferroni post hoc analysis revealed a significant difference between persons with severe and mild haemophilia (p = .037).

Regarding the lean mass, the analysis of the subregions (four extremities) revealed, that there is a significant difference between the severity phenotypes in the legs and the right arm (see Table 3). Bonferroni post hoc analyses revealed a significant difference considering the right arm between persons with severe compared to persons with mild haemophilia (p = .035) as well as persons with moderate haemophilia (p = .020). There was no significant difference considering the left arm (p = .058). Appendicular fat mass is not significantly different between severity phenotypes (p > .05).

TABLE 3.

Lean‐ and fat‐mass of subregions within different severity phenotypes.

Variable	Total (n = 201)	Severe (n = 116)	Moderate (n = 41)	Mild (n = 44)	p‐value
Arm left fat (g) M ± SD Median [IQ25–75]	1535 ± 613 1431 [1120–1857]	1566 ± 674 1467 [1081–1892]	1515 ± 572 1401 [1140–1790]	1428 ± 475 1375 [1050–1832]	.737
Arm left lean (g) M ± SD Median [IQ25–75]	3355 ± 638 3305 [2932–3775]	3273 ± 660 3167 [2854–3748]	3475 ± 596 3415 [3106–3861]	3460 ± 594 3419 [3011–3774]	.058
Arm right fat (g) M ± SD Median [IQ25–75]	1548 ± 580 1443 [1148–1826]	1556 ± 639 1409 [1092–1945]	1523 ± 527 1420 [1123–1709]	1515 ± 471 1484 [1151–1865]	.995
Arm right lean (g) M ± SD Median [IQ25–75]	3596 ± 621 3562 [3169–4006]	3479 ± 636 3409 [3102–3844]	3757 ± 573 a 3731 [3459–4119]	3755 ± 568 a 3786 [3267–4072]	.004
Leg left fat (g) M ± SD Median [IQ25–75]	3929 ± 1606 3703 [2845–4445]	3992 ± 1752 3786 [2775–4625]	3857 ± 1496 3661 [2901–4308]	3717 ± 1297 3619 [2791–4357]	.752
Leg left lean (g) M ± SD Median [IQ25–75]	9053 ± 1691 8951 [7959–10114]	8836 ± 1907 8651 [7503–10029]	9407 ± 1326 9286 [8311–10081]	9295 ± 1275 9149 [8493–10415]	.028
Leg right fat (g) M ± SD Median [IQ25–75]	3988 ± 1613 3804 [2947–4498]	4025 ± 1718 3816 [2922–4551]	4008 ± 1607 3740 [2994–4474]	3750 ± 1343 3581 [2762–4407]	.705
Leg right lean (g) M ± SD Median [IQ25–75]	9054 ± 1605 8966 [8028–10029]	8861 ± 1802 8645 [7501–10036]	9288 ± 1368 9163 [8437–9976]	9251 ± 1144 9116 [8441–10171]	.034

Note: Data displayed as mean ± SD and median [Interquartile range 25–75].

^a Indicates significant difference (Bonferroni corrected) to severe haemophilia: Arm right lean: severe‐mild: p = .035, severe‐moderate: p = .020.

To provide age‐related reference data, the cohort was divided into five different age groups (18–29 [n = 50], 30−39 [n = 47], 40−49 [n = 26], 50−59 [n = 43], 60−80 [n = 35]). It was found that fat‐related parameters such as fat/height², body fat percentage, VAT, android fat and android/gynoid ratio, respectively, significantly differ across the age groups (Table 4). Hereby, the difference is mainly observed between the youngest age group (18–29 years) compared to persons aged 40−49, 50−59 and 60−80 years, respectively. However, between the three older age groups only trends can be observed that indicate higher fat content with increasing age.

TABLE 4.

Overall body composition parameters within different age groups.

Variable	Age group 18−29 (n = 50)	Age group 30−39 (n = 47)	Age group 40−49 (n = 27)	Age group 50−59 (n = 42)	Age group 60−80 (n = 35)	p‐value
BMI (kg/m2) M ± SD Median [IQ25–75]	24.0 ± 2.9 24.0 [22.2–26.0]	26.9 ± 6.7 24.9 [23.4–28.6]	26.6 ± 4.1 26.8 [23.8–28.4]	27.1 ± 3.9 a 26.9 [24.8–30.2]	26.6 ± 3.1 a 26.1 [24.7–27.7]	<.001
Fat/height2 (kg/m2) M ± SD Median [IQ25–75]	6.2 ± 2.0 6.1 [4.4–7.3]	8.1 ± 4.3 6.9 [5.8–9.3]	8.5 ± 2.8 a 8.5 [6.0–9.8]	8.8 ± 6.0 a 7.8 [6.2–9.8]	8.5 ± 2.2 a 8.2 [7.1–9.0]	<.001
Lean/height2 (kg/m2) M ± SD Median [IQ25–75]	16.6 ± 1.9 16.6 [14.8–18.2]	17.2 ± 2.9 16.8 [15.6–18.3]	17.2 ± 1.9 17.2 [16.0–18.3]	19.2 ± 9.7 17.9 [16.3–19.2]	17.0 ± 1.6 16.9 [15.6–18.1]	.106
Body fat percentage (%) M ± SD Median [IQ25–75]	25.9 ± 6.0 26.7 [21.1–29.1]	29.5 ± 7.1 28.6 [25.5–33.1]	31.2 ± 5.3 a 31.5 [26.3–35.2]	29.5 ± 5.1 29.7 [26.1–34.1]	31.9 ± 5.0 a 31.1 [27.5–35.6]	<.001
Est. VAT mass (g) M ± SD Median [IQ25–75]	404 ± 150 370 [286–477]	667 ± 495 480 [375–830]	808 ± 317 a 729 [578–955]	868 ± 370b, a 769 [608–1140]	1092 ± 405 a , b 1034 [797–1255]	<.001
Android fat (g) M ± SD Median [IQ25–75]	1615 ± 775 1450 [1020–1946]	2505 ± 1932 1860 [1508–3074]	2896 ± 1246 a , b 2806 [1949–3516]	2761 ± 1078 a 2744 [2022–3441]	3001 ± 962 a , b 2814 [2404–3296]	<.001
Gynoid fat (g) M ± SD Median [IQ25–75]	3614 ± 1312 3538 [2641–4166]	4297 ± 2157 3852 [3249–4800]	4084 ± 1709 3617 [2870–4840]	3752 ± 1127 3860 [3021–4207]	3530 ± 935 3360 [2918–3964]	.280
Android/Gynoid Ratio M ± SD Median [IQ25–75]	0.4 ± 0.1 0.4 [0.3–0.5]	0.5 ± 0.1 a 0.5 [0.4–0.6]	0.7 ± 0.2 a , b 0.7 [0.6–0.8]	0.7 ± 0.2 a , b 0.7 [0.6–0.8]	0.8 ± 0.2 a , b 0.8 [0.7–1.0]	<.001

Note: Data displayed as mean ± SD and median [Interquartile range 25–75].

Abbreviation: Est., estimated; VAT, Visceral adipose tissue.

^a Indicates significant difference (Bonferroni corrected) to age group 18−29.

^b Indicates significant difference (Bonferroni corrected) to age group 30−39; Kruskal–Wallis Test was used: BMI: 18−29 to 50−59: p = .001; 18−29 to 60−80: p = .010; Fat/height²: 18−29 to 40−49: p = .005, 18−29 to 50−59: p = .001, 18−29 to 60−80: p < .001; Body fat percentage: 18−29 to 40−49: p = .008, 18−29 to 60−80 p < .001; Est. VAT: 18−29 to 30−39: p = .010, 18−29 to 40−49: p < .001, 18−29 to 50−59: p < .001, 18−29 to 60−80 p < .001, 30−39 to 50−59: p = .014; 30−39 to 60−80: p < .001; Android fat: 18−29 to 40−49: p < .001, 18−29 to 50−59: p < .001, 18−29 to 60−80 p < .001, 30−39 to 60−80: p = .010, Android/gynoid ratio: 18−29 to 30−39: p = .027 18−29 to 40−49: p < .001, 18−29 to 50−59: p < .001, 18−29 to 60−80 p < .001, 30−39 to 40−49: p = .004, 30−39 to 50−59: p < .001; 30−39 to 60−80: p < .001.

Going more into detail, lean and fat mass of the legs do not differ between the different age groups (see Table 5). But fat mass of the arms differs significantly between the youngest age group (18–29) and the groups of persons aged 30 and older, indicating a linear increase of arm fat with increased age. However, post hoc analysis showed that there is no significant difference within age groups 40−49 to 50−59 or 60−80.

TABLE 5.

Lean‐ and fat mass of subregions within different age groups.

Variable	Age group 18−29 (n = 50)	Age group 30−39 (n = 47)	Age group 40−49 (n = 27)	Age group 50−59 (n = 42)	Age group 60−80 (n = 35)	p‐value
Arm left fat (g) M ± SD Median [IQ25–75]	1188 ± 425 1065 [860–1500]	1606 ± 880 1335 [1126–1890]	1655 ± 498 a 1693 [1266–1878]	1590 ± 473 a 1588 [1228–1875]	1750 ± 466 a 1720 [1401–2126]	<.001
Arm left lean (g) M ± SD Median [IQ25–75]	3352 ± 606 3273 [2895–3797]	3426 ± 659 3392 [3017–3864]	3331 ± 615 3275 [2987–3640]	3502 ± 622 3433 [3007–3835]	3106 ± 651 3118 [2686–3611]	.169
Arm right fat (g) M ± SD Median [IQ25–75]	1289 ± 468 1231 [918–1533]	1590 ± 851 1359 [1088–1774]	1658 ± 465 a 1595 [1280–2029]	1585 ± 435 a 1596 [1226–1939]	1716 ± 397 a 1677 [1443–2017]	<.001
Arm right lean (g) M ± SD Median [IQ25–75]	3641 ± 624 3490 [3224–4030]	3681 ± 557 3603 [3287–4101]	3693 ± 631 3623 [3158–4208]	3670 ± 546 3644 [3229–3894]	3256 ± 693 3152 [2565–3855]	.052
Leg left fat (g) M ± SD Median [IQ25–75]	3747 ± 1436 3610 [2646–4288]	4347 ± 2140 3945 [3122–4939]	4127 ± 1836 3723 [2583–4824]	3815 ± 1253 3946 [3096–4341]	3503 ± 976 3454 [2686–3981]	.379
Leg left lean (g) M ± SD Median [IQ25–75]	9257 ± 1531 9349 [7834–10429]	9292 ± 1900 9061 [8078–10113]	9148 ± 1524 9321 [8029–10031]	9052 ± 1561 8979 [8151–10235]	8363 ± 1790 8510 [7284–9946]	.164
Leg right fat (g) M ± SD Median [IQ25–75]	3799 ± 1488 3433 [2784–4392]	4435 ± 2119 4055 [3174–5092]	4222 ± 1832 3835 [2969–5106]	3780 ± 1216 3817 [3098–4300]	3562 ± 1013 3565 [2741–4336]	.296
Leg right lean (g) M ± SD Median [IQ25–75]	9232 ± 1429 9318 [8211–10435]	9239 ± 1858 9006 [8299–9678]	9041 ± 1727 9033 [7792–10280]	9182 ± 1357 8948 [8193–10115]	8407 ± 1576 8062 [7447–9674]	.100

Note: Data displayed as mean ± SD and median [Interquartile range 25–75].

Kruskal–Wallis Test was used.

Abbreviations: Est., estimated; VAT, Visceral adipose tissue.

^a Indicates significant difference (Bonferroni corrected) to age group 18−29: Arm left fat: 18−29 to 40−49: p = .001, 18−29 to 50−59: p = .002, 18−29 to 60−80 p < .001; Arm right fat: 18−29 to 40−49: p = .008, 18−29 to 50−59: p = .020, 18−29 to 60−80 p < .001.

4. DISCUSSION

This is the first study investigating a whole body DXA analysis in PwH. The focus was to differentiate between the haemophilia severities. It has been hypothesized that persons with severe haemophilia show altered body composition parameters, especially considering the lean mass, compared to persons with mild or moderate haemophilia. Primary results displayed in Tables 2 and 3, reveal that body fat distribution, estimated VAT, appendicular fat mass, and android/gynoid ratio do not differ between severity phenotypes. Emphasizing, the lean mass diverges over the severities, that is, persons with severe haemophilia show less lean mass compared to persons with moderate or mild haemophilia. Going more into detail, a differentiation between appendicular lean mass was conducted as especially persons with severe haemophilia are frequently affected by muscular atrophies due to haemarthroses and impaired joint functionality. Muscle atrophies imply a loss of lean mass, therefore it is understandable that persons with severe haemophilia suffer from a higher reduction of both appendicular lean mass and lean/height², although there was only a tendency in this respect on the left arm, which could be due to the fact that the frequency of bleeds is increased in the dominant (i.e., right) elbow. It is further assumed that significant difference is missed, given a high number of right‐handed subjects, leading to a systemic bias. However, the presence of muscular atrophy as a frequent comorbidity might be a key relevant factor for lower levels of lean mass. The whole muscle surrounding the affected joint, for example, musculus gastrocnemius, quadriceps muscle, the biceps brachii or triceps brachii degrade due to a deficiency of activation when trying to avoid pain. ²⁷ , ²⁸ It is known that overall contracting muscle mass mediates among others the myokine IL‐6 for instance, which positively affects metabolic aspects, as it increases insulin sensitivity and fat oxidation. ²⁹ Research further showed that immobilization causes a reduction of the volume of the musculus quadriceps as well as a decrease of whole‐body insulin sensitivity. ³⁰ However, it is still a matter of research to what extent local atrophies, which are present in PwH, play a role on the metabolism, though it can be assumed that there are the above‐mentioned alterations in PwH affected by atrophies.

It needs to be emphasized that reference data for body composition parameters of the general European population are limited. ³¹ There are two recent studies from Norway and Austria, respectively, providing reference data using the GE Lunar device, which can be used for comparison. ³² , ³³ Considering the BMI, the present data show an increased BMI (>25) in 56.2% of PwH, which corresponds to the male European prevalence of 54.4%, ³⁴ highlighting that both values are notably high. Prior research suggests that the presence of muscle atrophy is associated with lower BMI. ³⁵ This indicates that the BMI can be biased in PwH and might even be higher, respectively, to the high prevalence of atrophies within this study cohort. ⁸ , ³⁵ Considering the lean mass, the general Caucasian male population shows a lean/height² of 17.8 ± 1.8 kg/m², increasing with age. ³³ A gradual increase of lean mass with age is also seen in the presented study cohort, though overall lean mass is lower, especially in young PwH (lean/height² of PwH aged 18−29: 16.6 ± 1.9 kg/m²; reference data 17.3 ± 1.8 kg/m²). ³³

Regarding body fat distribution, reference data reveal a mean body fat percentage of 29.3 ± 7.3]% within the European male population, which equals the mean body fat percentage of the analysed PwH of 29.2 ± 6.2%. ³³ Taking a closer look at VAT, there is no consensus in literature for the respective reference values of the European population, but VAT strongly increases with age. ³³ , ³⁶ Reference data suggest a mean VAT of 424 ± 385 g in healthy subjects aged 18−29, which is similar, though even higher compared to the data of PwH (404 ± 150 g). In healthy subjects aged 40−80 (65.9 ± 9.1) years the mean VAT is 1660 ± 687 g, which is higher than in PwH (see Table 4). ³² , ³³ This difference might partially be explained by the fact that especially the older age groups of the reference data showed an increased BMI, leading to a systemic bias and a high standard deviation. ³³ , ³⁶ The low proportion of VAT in haemophilia should be investigated further.

In spite of that, the mean android/gynoid ratio of 0.6 ± 0.2 is similar compared to non‐haemophilic European reference data (ratio 0.7 ± 0.2). To the best of our knowledge, there are no previous studies on android/gynoid ratio in PwH, though data might be comparable to waist‐to‐hip‐ratio. Kennedy et al. revealed that in 66% (n = 35/53) of the PwH analysed, the waist‐to‐hip‐ratio is increased. ⁸ However, android fat is known to be an indicator for cardiovascular risk, while in contrast gynoid fat might be rather protective. ³³ , ³⁷ With regard to fat‐related parameters, the presented findings do not show any significant differences between the haemophilia severities, which is in in line previous findings. ⁸ Either way, in persons with high VAT and android fat, it is recommended to monitor individual vascular health in persons with high VAT and android fat. Physical activity in PwH should be promoted to oppose reduced lean mass, increased (android) fat distribution and the associated risks. The training schedule should therefore entail both, the training of isolated muscle groups persons with atrophies as well as aerobic exercising to reduce (android) fat distribution. ³⁸

Moreover, the body composition parameters were also displayed and analysed in regard to different age groups. Here, it can be seen that fat parameters, such as fat/height², total body fat percentage, VAT, android fat, and the android/gynoid ratio, respectively, increase with age. It has further been revealed, that the absolute amount of appendicular fat varies across age groups. There is less fat of the arms in young PwH compared to older PwH, which is consistent with previous findings. ³⁹ The fat mass of the legs does not differ between age groups in PwH. It was found to be highest in persons aged 40−59 and decreases again with advanced age. These findings are not consistent with previous literature but might be explained through high prevalence of atrophies in older PwH and the concomitant altered limb composition. ³⁹

However, an increase of body fat and VAT is physiologically linked to age, which has been stated in previous literature. ³³ Thus, the present findings are unsurprising, though it needs to be highlighted that the life expectancy of the haemophilic population has fortunately increased as treatment options have improved. ²⁸ Hence, a higher awareness on age‐related comorbidities implying the consequences of higher fat distribution is required, and should be considered with regard to individual cardiometabolic risk assessment.

4.1. Strengths and limitations

One major strength of this study is the methodology since DXA is considered as gold standard for body composition analyses. Therefore, this investigation provides reliable results, extending prior literature on body composition in PwH, and can be used as reference data on PwH in both future research and clinical routine. the sample size and age range of this study is large (n = 201, 18−79 years). Nevertheless, this study is not able to provide reference data for age groups differentiated by severities, which is because the subgroups are too small. As body composition diverges within ethnicities, it needs to be highlighted that this study cohort is of Caucasian ethnicity and can therefore only apply as reference in Caucasian populations. ⁴⁰ Being aware of the fact that the present data were gathered using Hologic DXA system, the transferability to other DXA systems might be limited. ³³ The difference within these systems further limits the degree of comparing the present PwH to European reference data. Moreover, effects of nutrition have not been investigated in this study. As nutrition has a major impact on body composition it should be investigated in future research.

5. CONCLUSION

This study provides reference data for body composition parameters in PwH determined by DXA in a representative sample. persons with severe haemophilia have significantly less lean mass than persons with moderate or mild haemophilia. This can be attributed mainly to a higher presence of muscle atrophy. PwH show low VAT mass, average body fat percentage, and android/gynoid ratio compared to Caucasian individuals without bleeding disorders. These fat‐related body composition parameters do not differ between the severity phenotypes. Age is directly linked to increased fat in PwH, but does not affect the amount of lean mass. As the haemophilic population ages due to better pharmacologic treatment options, awareness of age‐related comorbidities, including fat‐related comorbidities, should rise in haemophilia management.

AUTHOR CONTRIBUTIONS

P. Ransmann, M. Brühl and J. Hmida performed the data collection and analysis. G. Goldmann and J. Oldenburg supported with the recruitment process. T. Hagedorn, T. Hilberg, A.C. Strauss², A.C. Strauss⁴ and F. A. Schildberg contributed scientifically to the manuscript. T. Hilberg and A.C. Strauss² designed and supervised the project.

CONFLICT OF INTEREST STATEMENT

Pia Ransmann received speakers’ fees and travel support from Takeda Pharmaceuticals as well as travel support from Sobi. Marius Brühl has received travel support from Takeda and Sobi. Andreas Christian Strauss has received research funding from Bayer, Sobi and Takeda as well as consultancy, speaker's bureau, honoraria, scientific advisory board and travel expenses from Bayer, Biotest, CSL Behring, Novo Nordisk, Sobi and Takeda. Georg Goldmann has received advisory board and travel expenses from Bayer, BioMarin, Biotest, Chugai Pharmaceutical Co. Ltd, CSL Behring, Grifols, LFB, Novo Nordisk, Octapharma, Pfizer, F. Hoffmann‐La Roche Ltd, Sobi and Takeda. Johannes Oldenburg has received research funding from Bayer, Biotest, CSL Behring, Octapharma, Pfizer, Sobi and Takeda as well as consultancy, speakers bureau, honoraria, scientific advisory board and travel expenses from Bayer, Biogen Idec, BioMarin, Biotest, Chugai, CSL Behring, Freeline, Grifols, LFB, Novo Nordisk, Octapharma, Pfizer, Roche, Sanofi, Spark Therapeutics, Sobi and Takeda. Thorsten Hagedorn has received research funding from Biotest, Chugai, Intersero, Novo Nordisk, Roche, Sobi and Takeda as well as travel expenses, speaker or scientific advisory board honoraria from Bayer, Biotest, Chugai, Novo Nordisk, Pfizer, Roche, Sanofi, Sobi and Takeda

ETHICS STATEMENT

This study was conducted in accordance with the principles of good clinical and ethical practice and was approved by the local ethic committee (339/19). Along with the Declaration of Helsinki, all participants gave written informed consent.

Ransmann P, Brühl M, Hmida J, et al. Determination of body composition by dual x‐ray absorptiometry in persons with haemophilia. Haemophilia. 2024;30:1332–1340. 10.1111/hae.15091

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon request.