The effect of body size and composition on lumbar spine trabecular bone score in morphologically diverse subjects

Abstract

Aim

The trabecular bone score (TBS) is a tool for assessing bone quality and health. Current TBS algorithm corrects for body mass index (BMI), as a proxy of regional tissue thickness. However, this approach fails to consider BMI inaccuracies due to individual differences in body stature, composition and somatotype. This study investigated the relationship between TBS and body size and composition in subjects with a normal BMI, but with large morphological diversity in body fatness and height.

Methods

Young male subjects (n = 97; age 17.2±1.0 years), including ski jumpers (n = 25), volleyball players (n = 48) and non-athletes (controls n = 39), were recruited. The TBS was determined from L1-L4 dual-energy X-ray absorptiometry (DXA) scans using TBSiNsight software.

Results

TBS correlated negatively with height and tissue thickness in the L1-L4 area in ski jumpers (r = -0.516 and r = -0.529), volleyball players (r = -0.525 and r = -0.436), and the total group (r = -0.559 and r = -0.463), respectively. Multiple regression analyses revealed that height, L1-L4 soft tissue thickness, fat mass and muscle mass were significant determinants of TBS (R² = 0.587, p<0.001). L1-L4 soft tissue thickness explained 27% and height 14% of the TBS variance.

Conclusion

The negative association of TBS and both features suggests that a very low L1-L4 tissue thickness may lead to overestimation of the TBS, while tall stature may have the opposite effect. It seems that the utility of the TBS as a skeletal assessment tool in lean and/or tall young male subjects could be improved if tissues thickness in the lumbar spine area and stature instead of BMI were considered in the algorithm.

Introduction

The Trabecular Bone Score (TBS) is a textural parameter which provides an indirect evaluation of trabecular microarchitecture in the lumbar spine (L1-L4) [1]. Existing data indicate that a low L1-L4 TBS is associated with both a history of fracture and occurrence of new fractures in the future [2]. Because TBS is independent from bone mineral density (BMD) in bone fragility evaluation [3], it offers a valid and complementary tool to improve bone health assessment [1, 2] and thus, has been included in the WHO Fracture Risk Assessment Tool (FRAX) [4].

TBS is derived from pixel grey-level variation in the lumbar spine (2D) DXA image. The texture of the DXA images depends not only on different X-ray absorption characteristics of regional bone, but also on body size and composition. The presence of soft tissues leads to X-ray attenuation, affecting the detection of grey-level variations, which may eventually alter the TBS estimate [5]. To overcome this error, the TBS algorithm considers body mass index (BMI) as a surrogate of regional soft (mainly fat) tissue thickness. Nevertheless, there is evidence that BMI may not be a sufficiently specific proxy and its clinical limitations might affect the TBS estimation [5–7]. For example, BMI does not distinguish between excess fat and muscle mass, nor does it provide any information on body size or fat distribution. Consequently, BMI might misclassify lean individuals with highly developed muscle as being overweight or obese, which often occurs in athletes.

Interestingly, there is data indicating that not only BMI and regional soft tissue characteristics, but also body stature [6, 7], might lead to biased estimates of the TBS. To date, studies investigating the possible effect of individual indicators of body stature and composition on the TBS are limited, ambiguous, and target only non-physically active adults [8, 9] and children aged 10–17 years [10]; hence, more research is needed to understand this relationship more completely, thereby informing algorithm development to overcome this bias.

The study aim was to investigate the relationship between TBS and body size and composition in male subjects, with particular emphasis on soft tissue in the L1-L4 area. To test the influence of body size and structure, as potential confounders in computing a TBS, we targeted lean adolescent subjects characterized with a normal BMI, but of highly diverse height and body composition. These conditions were met by representatives of two sports; volleyball (representing a very tall, slender body shape) and ski jumping (representing a lean body and average stature). We broadly hypothesized that body stature (height) and composition (soft tissues) would affect TBS estimation.

Materials and methods

Subjects

Ninety-seven young male subjects, aged 15–19 years, were recruited for this study. This cohort consisted of 25 ski jumpers (SJ) and 48 volleyball players (VP) who were members of national youth teams, and 39 non-athletes who formed a control group. The presented results are part of a more comprehensive project implemented in a group of physically active and untrained male youth. The exclusion criteria were: obvious signs of acute or chronic illness, which may affect bone health, a long period of immobilization or bone fractures in the previous 12 months, BMI >37 kg/m² and <15 kg/m² [11, 12]. The basic characteristics of studied groups are presented in Table 1. Biological age was estimated via bone age, based on activity of bone alkaline phosphatase (BALP).

[Figure omitted. See PDF.]

This study was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association. The experimental protocol was approved by the Ethics Committee (KEBN-18-39-JM). Written informed consent was obtained from the subjects or their parents when the participant was under the age of consent.

Procedures

DXA measurements.

The DXA measurements were performed using the Lunar Prodigy DXA scanner (GE Healthcare Inc, Madison USA) and analysed using the GE Encore software (version 1.6). The DXA machine was calibrated at the start of each testing day using a quality assurance block and a spine phantom. The DXA scanner exhibited high reliability for the assessment of lumbar spine bone mineral content (BMC; CV = 0.6%), lumbar spine areal bone mineral density (aBMD; CV = 0.4%), and TBS (CV = 1.4%).

Lumbar spine BMD, BMC, and TBS measurements.

In all subjects, aBMD (g/cm²) and BMC(g) were measured for the L1-L4 vertebrae. Age-matched Z-scores for the L1-L4 area were calculated with GE Encore software, using age-appropriate reference values. Since DXA-derived measurements of bone density provide details for area and not volume, L1-L4 bone mineral apparent density (BMAD) was calculated using a validated formula: BMAD = (L1 BMC + L2 BMC + L3 BMC + L4 BMC) / (L1V + L2V + L3V + L4V) [12]. The volume of each vertebra was calculated as the anterior-posterior bone area of the respective vertebra, raised to the power of 1.5.

Trabecular bone microarchitecture in the L1-L4 was assessed indirectly using TBS iNsight (Osteo) software (Version 3.0.2, Medimaps, Pessac, France). When the Z-score values for BMD in adjacent vertebrae differed by more than 1 SD, the outlier result was excluded, and the mean values of BMD, Z-score and TBS were calculated from 3 vertebrae [13]. In adults, values ≥1.350 are classified as normal, whereas values from 1.200 to 1.350 are recognized as partially degraded, and values <1.200 represent fully degraded bone microarchitecture [14].

Body composition.

Body composition was measured in the morning (8.00–9.00 a.m.) using a bioelectrical impedance analysis (BIA) system (Tanita BC-420MA, Japan). Subjects were fasted, bladder voided, wearing only underwear, with all metal objects removed. The following body components were assessed: total fat mass (FM), fat-free mass (FFM), and muscle mass (MM), all expressed in kg and as a percentage. Data on L1-L4 tissue thickness (cm), as well as tissue fat percentage (%) in this region, were derived automatically from the DXA report.

Blood assays.

Blood was obtained from an antecubital vein at 8.00–9.00 am, under fasting conditions, at least 38 h after the last exercise bout. Blood samples were collected into tubes containing coagulation accelerator and serum separator before centrifugation (10 min at 2000 × g). The serum aliquots were stored at -20°C for further tests of bone-specific alkaline phosphatase (BALP) activity. Sample testing was performed using a commercial enzyme immunoassay (IDS-iSYS CrossLaps Immunodiagnostic Systems Ltd, UK). The intra- and inter-assay CVs were 4.1% and 3.6%, respectively.

Anthropometric measurements.

Body height (Seca 285, Germany), body mass and fat (Tanita BC-420MA, Japan), as well as waist circumference (cm) were assessed. For the latter, five skinfolds (triceps, subscapular, iliac crest, supraspinale, medial calf) were taken using a Harpenden skinfold caliper. All measurements were taken by the same researcher according to the ISAK [15]. Somatotype was determined using the Heath-Carter anthropometric method [16]. BMI was calculated as weight (kg) divided by height squared(m²), and waist-to-height ratio (WHtR) as waist circumference (cm) divided by height (cm).

Statistical analyses.

Data distribution and homogeneity of variance were explored by the Shapiro-Wilk test and Levene’s test, respectively. Differences between groups were determined by a one-way analysis of variance with post-hoc comparisons using the Tukey test or Kruskal-Wallis test. Pearson correlations were used to explore associations between TBS and the different body size and composition variables.

To investigate which independent indices of body size and composition might predict the TBS, we performed a backwards stepwise multiple linear regression analysis on the pooled dataset (i.e., total group). To determine the relative importance of each predictor, the unique contribution of each variable was calculated using commonality analysis [17]. Alpha values of p<0.05 were considered statistically significant. Results are presented as means (±SD). Statistica 13.1 software (Dell Inc., Tulsa, OK, USA) was used for data analyses.

Results

The studied groups did not differ in terms of chronological age, BALP activity or, in the case of athletes, training experience (Table 1). According to our assumptions, the groups differed significantly (p<0.001) in terms of height and fatness (endomorphic). Moreover despite similarities in the slenderness (ectomorphic) and muscularity (mesomorphic) of both athlete groups, they differed in terms of body mass and BMI, stature and mean area of vertebrae in the L1-L4 region (p<0.001).

Bone tissue parameters

The TBS values were higher in ski jumpers (1.477±0.065; p<0.001) compared to both volleyball players (1.397±0.079; p<0.001) and controls (1.426±0.074; p<0.001). In contrast, volleyball players group had significantly higher (p<0.001) values of BMD and Z-score than the two other groups. However, after adjusting BMD for height (BMAD), the volleyball group differed significantly (p = 0.043) only from the control group (Table 2).

[Figure omitted. See PDF.]

Body size and composition indices

Tissue thickness in the L1-L4 area was significantly lower (p<0.001) in ski jumpers (17.3±1.07 cm) compared to the volleyball players (20.2±1.47 cm) and controls (20.7±1.07 cm) (Table 2). At the same time, significant between-group differences were found in the tissue fat content in this region (p<0.001) with tissue fat % being larger in controls (16.9±9.7%) than in volleyball group (6.2±1.7%) and SJ (5.0±1.0%). Relating to whole body composition, all groups differed (p<0.001) from each other in terms of fat mass percentage, as well as fat free mass and muscle mass expressed in kg and as a percentage. Body fat expressed in kg was significantly lower (p<0.001) in ski jumpers than volleyball players and controls.

Correlations of TBS with age and body size and composition

TBS correlated with height in both athletic groups (ski jumping r = -0.516; volleyball r = -0.525), and in the total group (r = -0.559) (Fig 1).

[Figure omitted. See PDF.]

Moreover, TBS correlated with mean vertebra area in all studied groups, both individually (ski jumpers r = -0.512; volleyball players r = -0.364 and controls r = -0.407) and in the total group (r = -0.522), as well as with L1-L4 regional soft tissue thickness in athletic males (ski jumpers r = -0.529; volleyball players r = -0.436) and the total group (r = -0.463) (Fig 2). Body mass correlated with TBS (r = -0.391), but only as a combined group.

[Figure omitted. See PDF.]

Relationships with body composition also emerged (Table 3). TBS correlated significantly with fat free mass and muscle mass in volleyball group (r = -0.330 and r = -0.329, respectively) and the total group (r = -0.468 and r = -0.469, respectively). Conversely, the corresponding correlations were not significant (fat free mass = -0.380, p<0.061; muscle mass r = -0.386, p<0.057) among ski jumpers (Table 3).

[Figure omitted. See PDF.]

The multiple regression results are presented in Table 4. Four of the body size and composition variables (height, fat mass, muscle mass and soft tissue thickness in L1-L4) explained in total 58.7% of TBS variation. L1-L4 soft tissue thickness made the largest unique contribution to the regression equation (explaining 27% of variation), followed by (in order from largest to smallest) height (14%), fat mass (11%), and muscle mass (6%).

[Figure omitted. See PDF.]

Discussion

The key study finding is that soft tissue thickness in the L1-L4 area and stature may affect the calculated indirect texture index of trabecular microarchitecture in the lumbar spine (TBS), at least in male subjects.

According to Silva et al. [18], bone size (which is a derivative of stature) does not affect the TBS, but recent studies suggest otherwise [6, 7]. For instance, a study in women with diverse height indicated that subjects with a lower body stature had higher TBS values than their taller counterparts [6]. Also, in male and female subjects with acromegaly, TBS values were lower compared to controls despite a comparable BMD [7]. Similar relationships were observed in Korean women and men with normal BMD T-score values [19]. In our study, TBS assessment, coming from male subjects diverse in height, also indicated lower than average TBS values in the tallest group (volleyball) compared to shorter athletes (ski jumping) and untrained peers (control). Additionally, values below 1.350, which in adults would be interpreted as degraded or partially degraded microarchitecture [14], were observed in nine volleyball players and six controls, but not in any ski jumpers. Furthermore, a negative relationship of TBS/height was observed both in athletes and the total sample, whereas a negative relationship of TBS/mean vertebra area was visible in all three groups, individually and collectively.

These relationships were verified via regression analyses, together with commonality testing across the total group; all of which identified height as one of the main determinants of TBS. That is, when height increases by 1 cm, the TBS would decrease by 0.006 units on average. To better understand the observed dependence, it is worth considering the technical aspects of DXA image texture processing aimed at extracting microstructural information. The TBS is an index based on quantification of variation in grey-level texture from one pixel to the adjacent pixels, i.e. the rate of grey-level differences between pixels at a specific distance [2, 11, 20]. Thus, a greater height, and hence vertebrae size, increases the number of pixels per vertebra. This means that, in taller subjects, the distance between two pixels represents a smaller relative fraction of the vertebra. Accordingly, and in line with our results, it might be speculated that a lower TBS in very tall individuals reflect erroneous calculation of this texture parameter, but not of worse bone microarchitecture in the lumbar spine. As such, TBS corrections for height seem justified, especially among very tall people, similarly to adjustment of aBMD towards lumbar spine bone mineral apparent density (BMAD), which in this study eliminated the volleyball and ski jumpers group’s difference. This issue requires longitudinal research, because it cannot be ruled out that the disturbed trabecular structure in tall young people could result from rapid bone growth, despite good mineralization. This assumption is supported by results of previous longitudinal studies conducted in young (approx. 13 years old) athletes [21], as well as cross-sectional studies on teenage girls and boys [10], showing that changes in stature are one of the positive predictors of the TBS.

The effect of soft tissue body components on the TBS is poorly understood. Some suggest that the TBS in the L1-L4, as opposed to BMD, is not influenced by the amount of regional soft tissue [22]. However, others reported that the TBS is negatively related to the amount of total fat mass, BMI and other variables measured in the L1-L4, such as trunk fat, visceral fat, trunk lean mass and waist circumference [8, 19, 23–27], as well as the tissue thickness in the measurement area [5, 25]. Our results in lean and very lean male subjects with normal BMI values confirm that TBS is negatively correlated with L1-L4 soft tissue thickness, while it does not depend on BMI, nor does it depend on the amount of fat deposited in the L1-L4 region. Interestingly, in controls, L1-L4 soft tissue thickness was similar to that observed in volleyball players, though it was not associated with TBS. Population differences in tissue composition in the measured region (i.e., a higher percentage of fat and thus a lower lean tissue in controls) offers one explanation. These observations suggest that regional lean tissue, rather than fat content, in the L1-L4 area might affect the TBS. On the other hand, regression analysis identified fat mass as one of four significant predictors of TBS. Previous results [5] suggest that fat and lean tissues might have independent effects on DXA images, due to different absorption characteristics. It is possible that, in the studied athletes, the percentage of fat in the L1-L4 area was so low that the soft tissue thickness in this area became the dominating factor. The influence of individual soft tissues (i.e., fat and lean soft tissue) on the TBS, particularly in the lumbar spine region, not only in male but also female subjects remains to be clarified.

It is also worth emphasizing that, in the literature, only increasing regional soft tissue thickness in L1-L4 area was addressed as a source of potential measurement error (i.e. lower TBS values in obese subjects), while so far, no study has addressed the influence of very low or decreasing soft tissue thickness in this area on the TBS. Meanwhile, among the very slim (lean) ski jumpers, a reduction in L1-L4 soft tissue thickness was linked to higher TBS values, which implies that a minimal amount of tissue in the measurement area can have an opposite effect, leading to overestimation of the TBS. Therefore, it cannot be excluded that the relatively high TBS observed in ski jumping group might reflect a generally thin soft tissue layer.

The present results are in line with suggestions by Kong et al. and Shevroja et al. that using only BMI, as a proxy for the L1-L4 soft tissue thickness, is the weakest component of the current version (3.0.2) of TBS software [1, 25]. This issue is especially pertinent to young subjects, due to the association with changes in body composition, bones, and muscle mass [17]. In the case of physically active populations, consideration of tissue composition in the target area is also justified because, in most of them, BMI does not reflect adiposity due to the larger amount of lean tissue. Therefore, our results suggest the need to modify the current TBS algorithm to account for regional soft tissue thickness [17]. Such a correction appears beneficial due to potentially falsely increased TBS in very lean and slender male subjects. Because the soft tissue thickness in the L1-L4 area is an easily accessible parameter derived from the lumbar spine DXA scan itself, at least in the case of densitometers from two popular manufacturers (i.e. Hologic and GE). Future research concerning tissue analysis in the L1-L4 area in the TBS aspect seems reasonable. Some study limitations should be mentioned; for example, the relatively small sample of solely male subjects limits the generalizability of the reported findings. Moreover, the number of ski jumpers was inadequate to create two separate regression models for each athletic group. Additionally, no direct measure of biological maturity was performed in this study; however, bone age, based on bone alkaline phosphatase activity, provided an index of maturation. Despite the apparent morphological differences, all studied groups presented a similar level of bone development and were of similar chronological age. Furthermore, due to technical limitations of the DXA scanner, we were unable to employ whole body DXA measurement in very tall subjects. To properly analyse the relationships of TBS with body composition (based on the same indicators for all subjects) the BIA method was used to assess body composition. Another limitation is the cross-sectional nature of this work, which made it impossible to account for individual growth rates during puberty, including shifts in body composition, and any impact on the TBS.

In conclusion, among young male individuals with normal BMI, but atypical body build (lean and/or tall), TBS was correlated with body size and composition, mainly with L1-L4 soft tissue thickness, stature and mean vertebra area in lumbar spine, wherein the first two features are the main determinants of TBS variability. Our findings suggest that a very small amount of lean tissue in the L1-L4 area in young male subjects may lead to overestimation of the TBS, while a tall stature may have an opposite effect. Thus, TBS estimates using current software may be partly artefactual and less reflective of the actual status of trabecular bone. It seems that the utility of the TBS could be improved by considering L1-L4 soft tissue thickness as well as stature (or lumbar vertebrae size). There is a need for future research in, morphologically diverse groups, both males and females, to refine the current TBS algorithm and validate it against direct methods (micro-computer tomography or quantitative computer tomography) of trabecular bone structure measurement. Until then, considerable caution is necessary when interpreting TBS, at least in young lean and/or tall subjects.

Citation: Malczewska-Lenczowska J, Surała O, Sitkowski D, Szczepańska B, Zawadzki M (2023) The effect of body size and composition on lumbar spine trabecular bone score in morphologically diverse subjects. PLoS ONE 18(7): e0287330. https://doi.org/10.1371/journal.pone.0287330

About the Authors:

Jadwiga Malczewska-Lenczowska

Roles: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Visualization, Writing – original draft

Affiliation: Department of Nutrition Physiology and Dietetics, Institute of Sport–National Research Institute, Warsaw, Poland

Olga Surała

Roles: Investigation, Methodology, Visualization, Writing – original draft

E-mail: [email protected]

Affiliation: Department of Nutrition Physiology and Dietetics, Institute of Sport–National Research Institute, Warsaw, Poland

ORICD: https://orcid.org/0000-0003-0419-9435

Dariusz Sitkowski

Roles: Software, Visualization, Writing – original draft

Affiliation: Department of Physiology, Institute of Sport–National Research Institute, Warsaw, Poland

Beata Szczepańska

Roles: Investigation, Writing – original draft

Affiliation: Department of Nutrition Physiology and Dietetics, Institute of Sport–National Research Institute, Warsaw, Poland

Maciej Zawadzki

Roles: Investigation, Writing – original draft

Affiliation: High School of Rehabilitation, Warsaw, Poland

References

1. Kong SH, Hong N, Kim J-W, Kim DY, Kim JH. Application of the Trabecular Bone Score in Clinical Practice. J bone Metab. 2021;28: 101–113. pmid:34130362

2. Harvey NC, Glüer CC, Binkley N, McCloskey E V, Brandi M-L, Cooper C, et al. Trabecular bone score (TBS) as a new complementary approach for osteoporosis evaluation in clinical practice. Bone. 2015/05/16. 2015;78: 216–224. pmid:25988660

3. Hans D, Goertzen AL, Krieg M-A, Leslie WD. Bone microarchitecture assessed by TBS predicts osteoporotic fractures independent of bone density: the Manitoba study. J bone Miner Res Off J Am Soc Bone Miner Res. 2011;26: 2762–2769. pmid:21887701

4. Krohn K, Schwartz EN, Chung Y-S, Lewiecki EM. Dual-energy X-ray Absorptiometry Monitoring with Trabecular Bone Score: 2019 ISCD Official Position. J Clin Densitom Off J Int Soc Clin Densitom. 2019;22: 501–505. pmid:31383412

5. Amnuaywattakorn S, Sritara C, Utamakul C, Chamroonrat W, Kositwattanarerk A, Thamnirat K, et al. Simulated increased soft tissue thickness artefactually decreases trabecular bone score: a phantom study. BMC Musculoskelet Disord. 2016;17: 17. pmid:26757709

6. Alvarenga PPM, Silva BC, Diniz MP, Leite MB, da Silva CAM, de Cássia Mendes Eleutério J, et al. Trabecular bone score: a useful clinical tool for the evaluation of skeletal health in women of short stature. Endocrine. 2019;66: 398–404. pmid:31302886

7. Hong AR, Kim JH, Kim SW, Kim SY, Shin CS. Trabecular bone score as a skeletal fragility index in acromegaly patients. Osteoporos Int. 2016;27: 1123–1129. pmid:26446771

8. Looker AC, Sarafrazi Isfahani N, Fan B, Shepherd JA. Trabecular bone scores and lumbar spine bone mineral density of US adults: comparison of relationships with demographic and body size variables. Osteoporos Int a J Establ as result Coop between Eur Found Osteoporos Natl Osteoporos Found USA. 2016;27: 2467–2475. pmid:26952009

9. Locquet M, Beaudart C, Durieux N, Reginster J-Y, Bruyère O. Relationship between the changes over time of bone mass and muscle health in children and adults: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2019;20: 429. pmid:31521141

10. Shawwa K, Arabi A, Nabulsi M, Maalouf J, Salamoun M, Choucair M, et al. Predictors of trabecular bone score in school children. Osteoporos Int. 2016;27: 703–710. pmid:26323328

11. Silva BC, Bilezikian JP. Trabecular bone score: perspectives of an imaging technology coming of age. Arq Bras Endocrinol Metabol. 2014;58: 493–503. pmid:25166040

12. Kindler JM, Lappe JM, Gilsanz V, Oberfield S, Shepherd JA, Kelly A, et al. Lumbar Spine Bone Mineral Apparent Density in Children: Results From the Bone Mineral Density in Childhood Study. J Clin Endocrinol Metab. 2019;104: 1283–1292. pmid:30265344

13. ISCD. Indications for Bone Mineral Density (BMD) Testing. 2019. Available: https://iscd.org/learn/official-positions/adult-positions/

14. Silva BC, Leslie WD, Resch H, Lamy O, Lesnyak O, Binkley N, et al. Trabecular bone score: a noninvasive analytical method based upon the DXA image. J Bone Miner Res. 2014;29: 518–530. pmid:24443324

15. Esparza-Ros F, Vaquero-Cristobal R, Marfell-Jones M. International Standards for anthropometric assessment. International Society for the Advancement of Kinanthropometry (ISAK), Spain; 2019.

16. Carter JEL, Heath BH. Somatotyping -development and applications. Cambridge University Press; 2003.

17. Majumdar S, Keavney TM, Rio L Del, Semler O, Winzenrieth R. What’s Next in the Field of Bone Health in Pediatrics? Research Considerations. Bone Health Assessment in Pediatrics. Springer; 2016. pp. 285–313.

18. Silva BC, Boutroy S, Zhang C, McMahon DJ, Zhou B, Wang J, et al. Trabecular bone score (TBS)—a novel method to evaluate bone microarchitectural texture in patients with primary hyperparathyroidism. J Clin Endocrinol Metab. 2013;98: 1963–1970. pmid:23526463

19. Kim Y-S, Han J-J, Lee J, Choi HS, Kim JH, Lee T. The correlation between bone mineral density/trabecular bone score and body mass index, height, and weight. Osteoporos Sarcopenia. 2017;3: 98–103. pmid:30775511

20. Winzenrieth R, Michelet F, Hans D. Three-dimensional (3D) microarchitecture correlations with 2D projection image gray-level variations assessed by trabecular bone score using high-resolution computed tomographic acquisitions: effects of resolution and noise. J Clin Densitom Off J Int Soc Clin Densitom. 2013;16: 287–296. pmid:22749406

21. Ubago-Guisado E, Vlachopoulos D, Fatouros IG, Deli CK, Leontsini D, Moreno LA, et al. Longitudinal determinants of 12-month changes on bone health in adolescent male athletes. Arch Osteoporos. 2018;13: 106. pmid:30306385

22. Schacter GI, Leslie WD, Majumdar SR, Morin SN, Lix LM, Hans D. Clinical performance of an updated trabecular bone score (TBS) algorithm in men and women: the Manitoba BMD cohort. Osteoporos Int a J Establ as result Coop between Eur Found Osteoporos Natl Osteoporos Found USA. 2017;28: 3199–3203. pmid:28733715

23. Messina C, Piodi LP, Rinaudo L, Emili I, Porro F, Buonomenna C, et al. Bone strain index reproducibility and soft tissue thickness influence: a dual x-ray photon absorptiometry phantom study. Eur Radiol Exp. 2019;3: 33. pmid:31410624

24. Leslie WD, Aubry-Rozier B, Lamy O, Hans D. TBS (trabecular bone score) and diabetes-related fracture risk. J Clin Endocrinol Metab. 2013;98: 602–609. pmid:23341489

25. Shevroja E, Aubry-Rozier B, Hans G, Gonzalez- Rodriguez E, Stoll D, Lamy O, et al. Clinical Performance of the Updated Trabecular Bone Score (TBS) Algorithm, Which Accounts for the Soft Tissue Thickness: The OsteoLaus Study. J bone Miner Res Off J Am Soc Bone Miner Res. 2019;34: 2229–2237. pmid:31419331

26. Bazzocchi A, Ponti F, Diano D, Amadori M, Albisinni U, Battista G, et al. Trabecular bone score in healthy ageing. Br J Radiol. 2015;88: 20140865. pmid:26148778

27. Lv S, Zhang A, Di W, Sheng Y, Cheng P, Qi H, et al. Assessment of Fat distribution and Bone quality with Trabecular Bone Score (TBS) in Healthy Chinese Men. Sci Rep. 2016;6: 24935. pmid:27112305