Epidemiology and Preventive Medicine, Sackler Faculty of Medicine, Tel-Aviv University

Transcription

1 Bone age assessment by a quantitative sonometer Short title: Bone age by ultrasound Marianna Rachmiel, MD 1,2, Larisa Naugolni, MD 1, Kineret Mazor-Aronovitch, MD 2,3, Amnon Levin 4, Nira Koren-Morag, PhD 5, Tzvi Bistritzer, MD 1,2. Ze ev Hochberg, MD PhD 6 1 Pediatric Endocrinology Clinic, Assaf Haroffeh Medical Center, Zerifin Israel, 2 Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel, 3 Pediatric Endocrinology Unit, Shebba Medical Center, Israel, 4 SonicBone Medical Company, Rishon Lezion, Israel, 5 Department of Epidemiology and Preventive Medicine, Sackler Faculty of Medicine, Tel-Aviv University 6 Pediatric Endocrinology, the Technion Israel Institute of Technology, School of Medicine, Haifa. Israel Corresponding author: Marianna Rachmiel, Pediatric Diabetes Service, Division of pediatrics, Assaf-Harofeh Medical Center, Zerifin, 70300, Israel. Tel: , Fax: , E.mail:rmarianna@gmail.com Key terms: short children, skeletal maturation, x-ray, bone age assessment, ultrasound, SonicBone, sonometer Conflict of Interest: The study was funded by SonicBone (Rishon Lezion, Israel). AL is an employee of SonicBone and ZH is an independent consultant Words count:

2 Abstract Objective: Bone maturation is currently assessed by repeated radiography using eye-balling subjective readings, or automated comparison of the characteristics of hand and wrist bones. The aim of the study was to evaluate the concordance and reproducibility of a new portable device (SonicBone's BAUS TM ) that utilizes quantitative ultrasonographic (US) technology measuring the speed of sound (SOS) of US waves propagating along measured bone, and to compare it to the current irradiating methods of bone age (BA) assessment (Greulich and Pyle (GP), Tanner- Whitehouse 3 (TW3)). Methods: The study population included 150 participants, 76 males, age years, attending an endocrine clinic. X-ray scans were evaluated independently by 4 observers. Separate readings for wrist, carpal and phalanx were averaged for child's BA assessment. Data from 100 subjects were utilized to assess the correlation between the speed-of-sound (SOS) and attenuation (ATN) parameters against the manual BA by GP method, and to establish a conversion equation for BA assessment by SonicBone. Fifty participants were assigned to assess validity. BA assessment by SonicBone was correlated for manual GP, automated GP and TW3 methods. Results: The coefficient of determination, R 2,for the conversion equation including gender, SOS and ATN was 0.80 for manual GP, 0.81 for automated GP and 0.82 for automated TW3 (p<0.001 for all). Pearson correlation between Sonographic BA and manual GP, automated GP and automated TW3 demonstrated significant validity, r=0.89, r=0.91, r=0.91 (p<0.001 for all), respectively. Conclusion: BA assessment by SonicBone is comparable to the assessment by all three X-ray based methods: manual GP, automated GP and TW3 2

3 . Introduction Skeletal maturity assessment, defined also as bone age (BA), is frequently used for evaluating growth and puberty in children and adolescents. It is recommended as part of the routine clinical care workup of a child with short or tall stature, precocious and delayed puberty, and more 1, 2. Repeated BA assessments are an important clinical tool utilized during the follow-up of such patients, especially when treated by growth and puberty-related interventions 1, 2. BA is currently assessed by radiography of the hand, using eye-balling or automated comparison of the shape and size of the wrist and hand bones to a standard series of representative radiographic films of hands according to the 1959 Radiographic Atlas of Skeletal Development by Greulich and Pyle (GP) 1, 3, or to the scoring method by Tanner and Whitehouse, currently in its 2001 third edition TW3 1, 4, 5. To address the disadvantage of repeated irradiation, the need for specialized radiation centers, heavy equipment and subjective reading 1, 6-8 a new device, SonicBone (SB) (Rishon Lezion, Israel) was developed. SB utilizes a quantitative ultrasonographic (QUS) technology of ultrasonic (US) waves, propagating along a measured bone distance 9, 10. Here, we report the validity and reproducibility of BA by SB, as compared to eye-balling BA by the GP method and an automated reading by both the GP and TW3 methods. Patients and Methods: Study design: This was a cross sectional study. The study subjects (n=150, 76 males) were recruited consecutively in a pediatric endocrine clinic. All BAs performed by X-ray scans and QUS were conducted prior to data analysis. The participants were then randomized to an Analysis group (n=100, 40 males) and a 'Validation group (n=50, 27 males). Analysis 3

4 performed on data obtained from the Analysis group enabled to establish a conversion equation for BA assessment by SonicBone. Analysis performed on data obtained from the Validation group assessed the relationship between QUS against the manual GP method, automated GP method, and automated TW3 method. The study protocol was approved by the institutional review board and by the Helsinki Committee of the Israeli Ministry of Health, and registered at (NCT ). Written informed consent was obtained from each legal guardian, and the participant assented for the study. Study population: Patients ranging between 4-17 years of age were recruited from the Pediatric Endocrinology Clinic at Assaf Haroffeh Medical Center, Israel. Inclusion criteria included all patients who performed hand x-ray scan as part of their clinical care. Exclusion criteria included children with bone diseases or those who within the last year took medications which might change bone metabolism or mineralization (such as: high dose steroids, biphosphonates, high dose Vitamin D, calcitriol). BA assessment by the manual GP method: Hand x-ray scans were reviewed and assessed independently by four pediatric endocrinologists being blinded to each other s findings. They assigned a separate BA to the long bones (Radius and Ulna, the carpalsand the short bones, and the mean of the three readings was defined as the child s BA, as previously described 11, 12. The mean of four observers was defined as the child s BA by GP. Automated BA assessment: The images were analyzed using the BoneXpert version 2.1 automated method for BA determination (Visiana, Denmark), which determines both GP and TW3 BA 13, 14 4

5 Ultrasonic BA assessment: The SonicBone (SB) device (Rishon Lezion, Israel) is a small (50cm X 25cm X 25cm), portable, bone sonometer (Figure 1), which measures two parameters: a) speed of propagation through bone (speed-of-sound, SOS, m/sec) of inaudible high frequency waves of a short ultrasound pulse; and b) attenuation (ATN; the decay rate) of the sound wave by the bone as a function of the distance it travels between a transmitter probe and a receiver probe 15. The hand was measured by (Figure 1): 1- wrist (W), measuring SOS and ATN at the distal radius and ulna secondary ossification centers of the epiphyses; 2-metacarpals (MC), measuring SOS and ATN at the distal metacarpal epiphyses; and 3- phalange (P), measuring SOS and ATN along the bent proximal third phalanx shaft, growth plate and epiphysis. The average of those 3 readings was defined as the child s BA by SB. All ultrasonic examinations were conducted at the Pediatric Endocrinology Clinic, by trained personnel. The examiners were blinded to the clinical background and to the BA by GP or TW3. Each subject underwent two readings by two observers. Eight additional repeated readings were performed for 10 subjects, 5 boys and 5 girls, aged 6-16 to ensure reproducibility and precision positioning assessment. Statistical Analysis: Data was analyzed with SPSS software version 21.0 (SPSS Inc. Chicago, Il, USA). The estimation of within subject repeatability was calculated by a one way analysis of variance (ANOVA) model. The outcome measures of the study included the correlation and hypothesis testing of equality of BA by SB and BA by manual GP, automated GP and automated TW3. The immediate side effects were also examined, by counting the numbers of incidents, the nature of inconveniencies and other complaints. The first phase analysis, performed on all study population (n=150), using Pearson correlation coefficient, demonstrated, the linear relationship between BA by GP and measurements of SB, SOS and ATN parameters. The second phase analysis of 100 subjects ( Analysis group ) established a conversion equation for estimating BA by SB out of the multivariate linear regression coefficients involving gender and the SB parameters. These equations provided the best R² result. 5

6 The third phase analysis was performed on data collected from the Validation Group (n=50) using the equations generated at phase 2.The correlation between BA by X-ray and by SB was analyzed by the Pearson correlation coefficient. Comparison of the differences between the BA by SB and by X-ray based methods was conducted by the paired t test and further presented by 95% confidence intervals. Results The study population included 150 subjects (76 males), mean age years ( years), recruited between June 2011 and March At the time of research investigation they were diagnosed with short stature and failure to thrive (46%), growth hormone deficiency (9%), precocious or early puberty (23%) overweight and obesity (8%), normal and healthy (14%). The clinical, demographic and body composition characteristics of the analysis and validation groups were similar (Table 1). According to measurements conducted independently in the same environment by two examiners, the performance analysis of SB revealed high reproducibility and repeatability Upon performing 10 repeated readings, on 10 subjects, the percent of relative standard deviation (%RSD) for SOS were smaller than 0.73% for all the children, with a maximum standard deviation of 13.7%. The %RSD for ATN was less than 3.5% for all the children with a maximum standard deviation of 1.4%. The distribution of SOS and ATN measurements according to skeletal area (W, MC, P) in the study population (n=150) is presented in Figure 2. The SOS measurements ranged from m/sec and the ATN surrogate distance ranged from mm. In the first phase analysis we correlated between BA by manual and automated GP and automated TW3 against SOS and ATN from SonicBone in all study population (n=150). A 6

7 significant correlation for both SOS ( r 2 =0.68, 0.68, 0.69 for manual GP, automated GP and automated TW3, respectively) and ATN (r 2 =0.88, 0.88, 0.89 for e manual GP, automated GP and the automated TW3, respectively) was found. In phase 2, multiple linear regression analysis was utilized in order to estimate the BA by the three methods using gender, SOS and ATN. The linear regression coefficients are presented in Table 2. For all three methods, both SOS and ATA were significantly strong predictors for BA. SOS was an important and significant predictor for BA above and beyond the ATN; the coefficient of determination R 2 significantly increased using SOS. As much as 82% of the total variation in BA is explained by ATN and SOS. Table 2 also shows the un-standardized coefficient b that are used for predicting future outcomes, and the standardized coefficients beta that were used to evaluate the relative strength of the relationship to BA. In the third phase analysis, the only data from the Validation group was compared to the assessment of BA, as delivered by the device according to the conversion equation performed in the second phase (BA by SB), to the three hand X-ray based methods. The results demonstrate a significant correlation between BA by SB and the BA by manual GP (r=0.89, P<0.001), automated GP (r=0.91, P<0.001) and automated TW3 (r=0.91, P<0.001). There were no clinical or statistical significance differences between methods, using paired comparisons (P=0.887; Table 3). Discussion In the dialectics of human anthropology and auxology, BA is an expression of the skeletal maturity of a child. Inferring from bone maturity, the clinician contemplate diagnostic considerations and evaluate height prediction, and would recommend sport activity and dancing, 7

8 or the timing for orthodontic procedures and orthopedic surgery. This is accomplished in a variety of methods, all of which utilize X-ray technology and compare a given film to various standards, followed by designation of a BA. The problematical utilization of repeated X-ray evaluations is well-appreciated. Here, we present the applicability of radiation-free BA assessment by QUS and its concordance with currently employed BA valuation by X-ray-based methods. This is not the first attempt to apply QUS technology for BA assessment; other attempts failed to enter clinical practice. Castriota-Scanderbeg et al 17, 18 attempted to assess skeletal maturation by quantifying the cartilage overlying layer of the femoral head. They demonstrated a decrease in cartilage thickness with age. Yet, a comparison with the BA by GP showed poor agreement 18. Shimura et al, 14 and Khan et al, 19, 20 assessed skeletal maturation by SOS through a single site at the head of the ulna (similar to W site in the current study), that often differs from other bones, not testing the sites available by SB device. 2, 11, 12. The SB device provides three independent measurements of the Radius and Ulna epiphyses, of metacarpals and of phalanges, similar to clinical practice assessments of hand x-ray scans. While the sites assessed by SB and by the X-ray methods are not identical, we demonstrate a significant correlation between BA by SB and BA by GP and TW3 methods at each site separately and by the mean BA 11. The BA by SB result was generated by the equation for BA assessment by QUS, which was integrated in the device, according to the data retrieved from the analysis of 100 subjects, including SOS, ATN and the manual reading by the GP method. It was then validated in 50 subjects against both manual and automated GP reading as well as the automated TW3 methods showing a high performance of reliability and significant concordance. 8

9 The ultrasound technique used by SonicBone is the "through transmission technique", as described in Figure 1. An ultrasound wave is propagated perpendicularly through a medium containing soft tissue and bone, from transmitter to receiver. Two parameters are used in this method: The primary parameter is SOS (time of flight) of the US wave over the distance from transmitter to receiver. SOS correlates strongly with the structure and density of the bone. However, bone is attenuative and dispersive. The attenuation is seen in the change in amplitude of a travelling wave and in the values of the reflection coefficients. The dispersion contributes to the distortion of the wave, and we therefore included in the equation the attenuation the decay rate of a wave as it propagates through bone. We use the distance between transmitter and receiver as an Attenuation factor - ATN. In the current study, the new technique was compared with manual reading by the GP method and with automated reading using both PG and TW3 methods. The latter gave the best regression against BA by SB; as much as 82% of the total variation in BA, is explained by ATN and SOS. For all three methods, the ATN showed important contribution to the regression, yet, in all three the SOS was important and significant predictor for bone age above and beyond the ATN. Thus, the measurements of SB device are all hand-area inclusive (W, MC, P), objective, and of physiological agreement to the goal of bone maturation assessment (SOS, ATN), offering a possible alternative to the present radiation based mostly subjective GP, and TW3 methods. As BA is an essential measurement procedure for pediatric Endocrinology physicians and quite often must be repeated over time, the SonicBone offers an important advantage over the current methods, with no side effects and with objective readings by a system accessible at the clinician office. The current report does not provide reference or standard for BA by chronological age for the QUS method. This is currently under development using a normal population distribution for all ages according to gender. 9

10 In summary, the radiation-free assessment of BA by the SonicBone device in pediatric population attending endocrinology clinics was found to be highly reproducible and comparable to the BA assessed by hand X-rays based methods, both subjective and objective methods. Author Contribution: MR, LN, KMA, AL, NKM, TB and ZH, made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data, took part in drafting the article or revising it critically for important intellectual content, and gave their final approval of the version to be published. MR, ZB, LN, NKM and ZH take responsibility for the integrity of the work as a whole, from inception to published article. References 1 Martin, D.D., Wit, J.M., Hochberg, Z., Savendahl, L., van Rijn, R.R., Fricke, O., Cameron, N., Caliebe, J., Hertel, T., Kiepe, D., Albertsson-Wikland, K., Thodberg, H.H., Binder, G. & Ranke, M.B. (2011) The use of bone age in clinical practice - part 1. Horm Res Paediatr 76,

11 2 Martin, D.D., Wit, J.M., Hochberg, Z., van Rijn, R.R., Fricke, O., Werther, G., Cameron, N., Hertel, T., Wudy, S.A., Butler, G., Thodberg, H.H., Binder, G. & Ranke, M.B. (2011) The use of bone age in clinical practice - part 2. Horm Res Paediatr 76, Greulich W, P.S. (1959) Radiographic atlas of skeletal development of the hand and wrist. Stanford university press. 4 Tanner JM, W.R. (1937) The atlas of skeletal maturation. Mosby Company, St. Louis,. 5 Tanner JM, W.R., Cameron N, Marshall WA, Healy MJ, Goldstein H., ed. (1975) Assessment of skeletal maturity and prediction of adult height: TW2 method. Academic Press, New York. 6 Bull, R.K., Edwards, P.D., Kemp, P.M., Fry, S. & Hughes, I.A. (1999) Bone age assessment: a large scale comparison of the Greulich and Pyle, and Tanner and Whitehouse (TW2) methods. Arch Dis Child 81, Johnson, G.F., Dorst, J.P., Kuhn, J.P., Roche, A.F. & Davila, G.H. (1973) Reliability of skeletal age assessments. Am J Roentgenol Radium Ther Nucl Med 118, Lynnerup, N., Belard, E., Buch-Olsen, K., Sejrsen, B. & Damgaard-Pedersen, K. (2008) Intra- and interobserver error of the Greulich-Pyle method as used on a Danish forensic sample. Forensic Sci Int 179, 242 e Specker, B.L. & Schoenau, E. (2005) Quantitative bone analysis in children: current methods and recommendations. J Pediatr 146, Zadik, Z., Price, D. & Diamond, G. (2003) Pediatric reference curves for multi-site quantitative ultrasound and its modulators. Osteoporos Int 14, Hochberg, Z. (2002) Endocrine control of skeletal maturation: Annotation to bone age readings. Karger Medical and Scientific Publishers. 11

12 12 Even, L., Andersson, B., Kristrom, B., Albertsson-Wikland, K. & Hochberg, Z. (2014) Role of growth hormone in enchondroplasia and chondral osteogenesis: evaluation by X- ray of the hand. Pediatr Res 76, Thodberg, H.H., Kreiborg, S., Juul, A. & Pedersen, K.D. (2009) The BoneXpert method for automated determination of skeletal maturity. IEEE Trans Med Imaging 28, Martin, D.D., Deusch, D., Schweizer, R., Binder, G., Thodberg, H.H. & Ranke, M.B. (2009) Clinical application of automated Greulich-Pyle bone age determination in children with short stature. Pediatr Radiol 39, Hosokawa, A. & Otani, T. (1997) Ultrasonic wave propagation in bovine cancellous bone. The Journal of the Acoustical Society of America 101, Jung, H. (2000) [The radiation risks from x-ray studies for age assessment in criminal proceedings]. Rofo 172, Castriota-Scanderbeg, A. & De Micheli, V. (1995) Ultrasound of femoral head cartilage: a new method of assessing bone age. Skeletal Radiol 24, Castriota-Scanderbeg, A., Sacco, M.C., Emberti-Gialloreti, L. & Fraracci, L. (1998) Skeletal age assessment in children and young adults: comparison between a newly developed sonographic method and conventional methods. Skeletal Radiol 27, Khan, K.M., Miller, B.S., Hoggard, E., Somani, A. & Sarafoglou, K. (2009) Application of ultrasound for bone age estimation in clinical practice. J Pediatr 154, Shimura N, K.S., Arisaka O, Imataka M, Sato K, Matsuura M (2005) Assessment of measurement of children's bone age ultrasonically with Sunlight BonAge. Clin Pediatr Endocrinol 14 (Suppl 24),

13 Table 1: Demographic, clinical and body composition parameters of study population, randomly divided into the analysis and the validation groups All Analysis Validation P Group Group value Number Gender (f) (51%) 23 (46%) 0.34 Pre-puberty* (39%) 17 (34%) 0.33 BMI SDS Age (years) Mean BA by GP - W (years) Mean BA by GP - CMC (years) Mean BA by GP - P (years) Mean BA by GP (all sites ) *Data is based on n=148, FTT- failure to thrive, GHD growth hormone deficiency. 13

14 Table 2: The regression coefficients, used in the conversion equation from the Analaysis Group Site BA-Manual GP BA-Automated GP BA- Automated TW3 Parameter Constant Gender SOS *(cm/sec) ATN **(mm) Constant Gender SOS *(cm/sec) ATN **(mm) Constant Gender SOS *(cm/sec) ATN **(mm) Coef.b SE Beta t P value R < < < < < < < < <0.001 SE= standard error, t=statistics value. *SOS- speed of sound, measurements ranges from 1,604 m/sec to 2,647 m/sec. ***ATN- attenuation, measurements ranges from 29.5 mm to 82.7 mm. 14

15 Table3: Comparison of BA by SonicBone to the BA by the three X-ray methods (in the validation group of subjects). Mean difference S.E. 95% C.I. P SB - manual GP SB - automated GP SB -automated TW

16 Figure legends: Figure 1: Bone age assessment by SonicBone bone sonometer at the third proximal phalange (A), the metacarpal (B) and the wrist (C). The device measures two parameters (D): the speed of propagation of inaudible high frequency waves of a short ultrasound pulse through bone; and the Attenuation/distance between a transmitter probe (T) and a receiver probe (R), located at the edges of the measured bone area for assessment of the attenuation factor. Figure 2: Quartiles distribution of SonicBone parameters of SOS and ATN according to the measured areas of the left hand, wrist (W), carpal (MC) and phalanx (P), in the whole study population (n=150). 1a. Speed of sound in m/sec (SOS). 1b. Attenuation/distance in mm (ATN). Lines within boxes indicate median; limits of boxes indicate 25th and 75th percentiles; circles represent outliers. 16

17 Figure 1 A B A B C C D 17

18 Figure 2a Figure 2b 18