Sarcopenia refers to the loss of skeletal muscle mass and strength, typically accompanying the normal aging process as well as in chronic conditions. Sarcopenia is a major cause of physical disability, poor quality of life, loss of independence, and death. Multiple modalities exist to evaluate muscle mass and nutritional status in clinical practice. While anthropometrics (such as weight, body mass index, mid‐arm circumference, and triceps skinfold thickness) and biochemical markers are readily clinically obtainable at the bedside, their values may be imprecise or confounded by disease‐related complications such as concurrent ascites, peripheral oedema and organomegaly seen in patients with end‐stage organ failure, malignancy, chronic inflammatory disease conditions, or post‐surgery. Specialized investigations exist to ascertain sarcopenia, including bioelectrical impedance and air‐displacement plethysmography. However, these are not routinely obtained clinical tests and require dedicated training and equipment thereby limiting the widest utilization. Measurement of psoas muscle area (PMA) from single slice cross‐sectional abdominal computed tomography (CT) images represents an easily accessible and quick method to assess the presence of sarcopenia. Amongst adults, multiple definitions of sarcopenia using PMA have been utilized, including distinct sex‐specific cut‐off values in healthy and diseased populations, as well as at different landmarks such as discrete lumbar (L3 and L4)disc levels, intervertebral lumbar (L3–4 and L4–5) disc levels, or the level of the umbilicus.
Sarcopenia is not a problem exclusive to the adults. Children with chronic medical illnesses experience nutritional deficiencies, physical deconditioning, and systemic inflammation that are potentially contributory to the unintentional muscle loss demonstrated in paediatric leukaemia, inflammatory bowel disease, type 2 diabetes, end‐stage liver disease, or intestinal failure. Sarcopenia has been evaluated in the paediatric population using dual energy X‐absorptiometry and abdominal CT imaging of the PMA at the L4 level. PMA reference values at various lumbar levels have been reported in the adult literature in acknowledgement that values may be impacted by refractory ascites and associated prominent lordosis, sacralization of the fourth lumbar vertebra, and lumbarization of the first sacral vertebra. However, reference values for adults cannot be extrapolated to growing children. Our goal was to generate sex‐specific and age‐specific PMA growth charts for intervertebral lumbar (L3–4 and L4–5) disc levels by reviewing clinically obtained cross‐sectional abdominal CT images for PMA values in a cohort of otherwise healthy children presenting after acute trauma to the emergency room of an academic health sciences centre.
We performed a cross‐sectional single‐centre study at the Hospital for Sick Children (SickKids) in Toronto, Canada. Patients were ascertained using the SickKids trauma database, a prospectively compiled database that houses clinical, biochemical, and radiological data from children presenting to the SickKids emergency room (ER). All children below 18 years of age presenting to the SickKids ER following acute trauma and requiring cross‐sectional abdominal CT imaging between January 2005 and December 2015 were eligible for inclusion. Children were excluded if their abdominal CT studies did not adequately scan the lumbar spine (L3–L5) region, or they had one or more of the following documented medical conditions: cerebral palsy, cardiac disease, malignancies, autoimmune diseases, diabetes mellitus, autistic spectrum disorders, thoracolumbar spinal trauma affecting paraspinal musculature, and chronic inflammatory disorders. The study was approved by the SickKids Research Ethics Board.
All medical charts were reviewed for relevant clinical information, including sex, age at CT (years), weight (kilogrammes), height (metres), and other anthropometric data available. Sex‐specific and age‐specific anthropometric z‐scores were calculated from the Center for Disease Control and Prevention (CDC) reference charts.
Abdominal CT images were retrieved from the SickKids Picture Archiving and Communication System (PACS). From axial CT images, L3–4 and L4–5 levels were identified following cross‐referencing on sagittal plane reconstructions. Landmark lumbar (L3 and L4) vertebrae were identified in the sagittal plane by counting down from the thoracic (T12) vertebra, which was assumed to be the most caudal rib‐bearing vertebra. In cases where the exact level was unclear, we identified the cervical C2 level (in available cases where CT of neck, chest, and abdomen was performed) and counted vertebral bodies downwards (Figure ). Alternatively, a chest X‐ray was reviewed to confirm 12 pairs of ribs were present, with the most caudal pair assumed to be T12. On the corresponding axial cross‐sectional images, we delineated bilateral PMA using a dedicated free‐hand region‐of‐interest measurement tool available on picture archiving and communication system at each of the intervertebral lumbar L3–4 and L4–5 levels, assessed in square millimetres (mm2) (Figure ). The total PMA (tPMA) was expressed as the sum of the right and left PMA in square millimetres (mm2) for each level. All PMA measurements were performed by one paediatric radiologist (HP). A random age‐stratified subset of CT images was measured bilaterally at both L3–4 and L4–5 levels by a second paediatric radiologist (GC) to ensure adequate interrater reliability. Both radiologists were blinded to clinical data.
Enlarge this image.Fig. 1. Axial and sagittal images. Lumbar levels were determined on (A) sagittal computed tomography images, enabling identification of the L3–4 and L4–5 intervertebral levels (yellow lines). (B) The axial images corresponding to these levels were then examined to determine the psoas muscle area (red lines).
Continuous variables were summarized using mean and standard deviation (SD). Categorical variables were summarized using frequencies and percentages. We used quantile regression with splines to model weight curves, which we compared with the CDC growth charts. A χ2 goodness of fit test was used to obtain the model with the best fit. We then used quantile regression with smoothed fractional polynomials to model the 5th, 10th, 25th, 50th, 75th, 90th, and 95th percentile curves for tPMA across patient ages between 1 and 16 years for both L4–5 and L3–4 levels after stratification by sex. Akaike's information criterion was used for model selection. We estimated the correlation between weight with tPMA (as measured at L3–4 and L4–5) stratified by sex using Pearson's correlation coefficients and interrater reliability for tPMA using intraclass correlation coefficients and Bland–Altman plots. Wilcoxon's paired signed‐rank test, overall and stratified by sex, was used to compare tPMA between lumbar levels. We estimated the correlation between lumbar levels using both Spearman's and Pearson's correlation coefficients and interrater reliability for tPMA using intraclass correlation coefficients and Bland–Altman plots. To allow a parametrization of the asymmetrical distribution of tPMA and derive z‐scores for an online tool, we determined standard deviations for half‐normal distributions above and below the median based on the median and 10th and 90th percentiles. Median, 10th, and 90th percentile for the online tool will therefore be identical to those published in this report. For remaining percentiles, there will be small differences between the online tool and published curves. The online tool is available as a Shiny web application built using R. The application allows users to input patient age (or date of birth and date of imaging), sex, lumbar level (L3–4 or L4–5), and tPMA (square millimetres, mm2). The age input is rounded to the nearest month and z‐scores are computed parametrically as described above. All analyses were carried out in R, version 3.5.0.
Review of the SickKids trauma database identified a total of 882 children requiring abdominal CT imaging at time of ER presentation between January 2005 and December 2015. Reasons for exclusion were established chronic medical condition (n = 33), acute lumbar spinal trauma (n = 5), and CT imaging not capturing the targeted L3 to L5 vertebral regions (n = 10). In addition, because of very small numbers of children <1 year (n = 10) and ≥16 years (n = 45), these patients were excluded as well. The final study cohort was comprised of 779 (36% female participants) children (Figure S1). The mean patient age was 9.5 years (SD 4.4). The age distributions for both sexes were similar (Figure S2). For male subjects, median (IQR) age was 9.91 (6.26–13.6) years. For female subjects, median (IQR) age was 9.86 (6.17–13.28) years.
Weight was normally distributed with a mean z‐score of 0.4 (SD 1.1) in the overall study cohort, 0.3 (SD 1.1) in girls, and 0.4 (SD 1.1) in boys. Figure displays the weight curves stratified by sex from 5th to 95th percentile for the study cohort in comparison with CDC growth charts. Children at the 50th percentile of our population were heavier than those at the 50th percentile of CDC growth charts, by an average of 3.3 kg in girls and 3.6 kg in boys (P < 0.0001). Height values were available in only 26 (3%) of our study cohort of 779 participants and therefore were not part of subsequent analyses.
Enlarge this image.Fig. 2. Weight distribution curves stratified by sex. Weight distribution curves (5th to 95th percentile) for (A) male and (B) female participants from the Hospital for SickKids (dashed lines) are compared with Center for Disease Control (CDC) growth charts (full lines). Weight curves differ significantly (P < 0.0001) for male and female participants, with this study's cohort being slightly heavier.
Growth charts for tPMA with 5th, 25th, 50th, 75th, and 95th percentile curves for boys and for girls at both lumbar levels are presented in Figure (L3–4) and Figure 4 (L4–5). At both L3–4 and L4–5 levels, tPMA values increased with patient age. Until the age of 7 years, tPMA values at both L3–4 and L4–5 levels were similar between sexes. From 8 years upwards, tPMA values were larger on average in boys than girls (P < 0.05). Values of tPMA at these percentiles in 6 months increments for patient ages 1 to 16 years are available. Values for tPMA at L4–5 were larger than those measured at L3–4 [mean difference 282 (95%CI 266 to 298) mm2, P < 0.05]. This difference was seen in both girls [mean difference 255 (95%CI 231to 280) mm2, P < 0.01] and boys [mean difference 298 (95% CI 278 to 318) mm2, P < 0.01]. The correlation between tPMA values at L3–4 and at L4–5 was high for girls (r = 0.95, P < 0.001) and boys (r = 0.98, P < 0.01). Psoas muscle configuration was subjectively rounder at L4–5 than L3–4. Figure shows the high correlation between weight and tPMA stratified by sex for boys (L3–4 r = 0.90; L4–5 r = 0.90) and girls (L3–4 r = 0.87; L4–5 r = 0.87). Interrater reliability between independent radiologists (coinvestigators HP & GC) in a random sample cohort (n = 53, 7%) was excellent, with intraclass correlation coefficients of 0.97 (95% CI 0.94 to 0.98) at the lumbar level L3–4 and 0.99 (95% CI 0.986 to 0.995) at L4–5. The Bland–Altman plots further support good agreement for tPMA measurements at L3–4 and L4–5 levels (Figure S3).
Enlarge this image.Fig. 3. Total psoas muscle area (tPMA) at lumbar level L3–4 showing 5th to 95th percentiles for (A) male and (B) female participants. tPMA values at L3–4 increases with patient age.
Enlarge this image.Fig. 4. Total psoas muscle area (tPMA) at lumbar level L4–5 showing 5th to 95th percentiles for (A) male and (B) female participants. tPMA values at L4–5 increases with patient age.
Enlarge this image.Fig. 5. Correlations of weight and total psoas muscle area (tPMA) at lumbar levels L3–4 and L4–5 stratified by sex. Weight and tPMA highly correlate for male and female participants at L3–4 and L4–5, respectively. (Pearson's correlation coefficient r = 0.90 for boys at L3–4 and L4–5 and r = 0.87 for girls at L3–4 and L4–5). Upper and lower dashed lines represent upper and lower boundaries of the 95% reference range.
The online tool is available at
Drawing on cross‐sectional CT imaging from almost 780 healthy children, this study proposes novel sex‐specific references values and percentile curves for total psoas muscle area at lumbar levels L3–4 and L4–5 for children between the ages of 1 and 16 years. Interobserver agreement was excellent at both lumbar levels, highlighting the feasibility of making these measurements. In addition, we created an online calculator that generates age‐specific and sex‐specific tPMA percentiles and z‐scores, further enhancing the practical application of this measure.
Sarcopenia traditionally denotes the loss of skeletal muscle mass and function that occurs with the normal aging process. Sarcopenia can also be a pathologic manifestation of disease states, including end‐stage organ disease, malignancies, and inflammatory diseases in children. The significance of sarcopenia derives from its association with adverse outcomes. For example, in adults with cirrhosis, sarcopenia, as assessed on CT, has been linked to impaired quality of life, hepatic decompensation, mortality in patients listed for liver transplantation (LT), increased length of hospitalization, infection, and higher health care costs. Sarcopenia is closely linked to frailty, a clinical syndrome reflecting increased vulnerability to medical stressors and adverse health outcomes. Frailty was first described as a clinical phenotype by Fried and colleagues in older adults. Numerous studies have since demonstrated frailty to be a prognostic marker for poorer outcomes in multiple adult populations.
The concepts of sarcopenia and frailty have been relatively neglected in the paediatric setting, yet both are likely highly relevant. Children are not exempt from the unintentional muscle loss because of nutritional deficiencies, physical deconditioning, and systemic inflammation present in chronic medical illnesses. Two paediatric pilot studies reported smaller tPMA values measured on CT in children with end‐stage liver disease as well as intestinal and kidney failure compared with matched healthy controls. Importantly, tPMA has been observed to have no association with traditional clinical anthropometric measures, highlighting the potential role of tPMA (and sarcopenia) as an independent biomarker of paediatric frailty. The results of this present study demonstrated a high correlation of weight and tPMA in healthy children. In a recent prospective multicentre cohort study of paediatric chronic liver disease, the assessment of frailty was adapted for paediatric application; school‐age children with end‐stage liver disease listed for LT were found to have significantly higher frailty scores than healthy controls, establishing the relevance of frailty in paediatric as well as adult populations. Because frailty assessments are difficult in children less than 6 years of age, the availability of an objective biomarker, such as sarcopenia using tPMA measurement, offers the opportunity to be inclusive of the youngest children who are particularly vulnerable to the sequelae and morbidities of end‐stage organ failure.
The lack of a consensus definition for sarcopenia is an important hindrance to the early identification of at‐risk patients who would benefit from early nutritional prehabilitation. In adults and adolescents, CT‐based tPMA values at multiple levels have been published. We elected to evaluate the psoas muscle at two intervertebral disc levels, L3–4 and L4–5 . Although a recent systematic review highlighted L3 to be a commonly reported level in adult patients to assess muscle mass on CT, anatomical variations may complicate reliable identification of borders of the muscle which often requires multiplanar reformation and can be time‐consuming. The psoas muscle becomes more round and distinct from other paraspinal muscles in lower lumbar levels (Chavhan, personal communication). When the vertebral body is used, there is variability in the exact anatomic definition of the level at which psoas muscle is measured. In contrast, the intervertebral disc space level is narrower/thinner and enable a more consistent level determination for measurement of the psoas. Several studies have reported measurements at L3–L4 and L4–L5. The feasibility of measuring tPMA at these two levels has been previously demonstrated in studies assessing paravertebral muscle area with other muscle groups and overall strength in adolescents and adults.
Harbaugh et al. recently published paediatric tPMA reference values at the mid‐L4 level derived from CT images in children presenting with acute appendicitis or acute traumatic injuries in Michigan. Our study provides reference range curves for tPMA values evaluated at intervertebral levels from CT scans in children. Additional differences between our study and that by Harbaugh et al. relate to the patient population. In Michigan, 76% of the population is White, followed by 14% Black and 5% Hispanic. As supported by the 2016 Canadian census, there is greater ethnic variation in Toronto (48% White, 13% South Asian, 11% Chinese, 9% Black, 6% Filipino, 3% Hispanic). In addition, the obesity rates amongst children In Canada are lower than those in USA. This is relevant as ethnicity and obesity have been shown to impact tPMA muscle mass values. The tPMA values derived using our Toronto cohort represent a more diverse population, and we propose to be potentially more broadly generalizable. We also expect that the availability of reference ranges at two landmarks (rather than one), coupled with an easy‐to‐use online calculator, will greatly facilitate the identification of at‐risk children. Such tools are needed because current conventional anthropometric measures, such as weight and skinfold thickness, are confounded by numerous factors, such as ascites or peripheral oedema and organomegaly, which are common in many chronic diseases. To this point, Suzuki et al. showed that in a paediatric leukaemia cohort, while some patients lost weight and others gained, all patients had a decrease in tPMA during induction treatment. tPMA is easily assessed using cross‐sectional CT imaging, which is frequently performed in the work‐up of children with end‐stage organ disease, malignancy, and other chronic conditions. This is critical as it means additional scan time is not needed solely for tPMA determination.
Our study has numerous strengths. They include the examination of a large, unbiased, and ethnically diverse paediatric cohort, with careful exclusion of children with medical conditions that might confound tPMA values. A subset of images was also read by two experienced radiologists to ensure reproducibility. Furthermore, we examined two levels, rather than just a single level. The L4–5 level may be the most clinically relevant and useful. This level has been referenced as the landmark for assessing visceral adipose tissue and thus one image at this level could be used for measurement of skeletal muscle (tPMA) and adipose tissue. The cross‐sectional psoas muscle shape at L4–5 was also a bit rounder than at L3–4 (where it was flatter), potentially facilitating more accurate contour drawing. Interrater reliability was slightly higher at L4–5 (intraclass correlation coefficient of 0.99). Lastly, an online application was subsequently developed to easily calculate age‐specific and sex‐specific z‐scores and percentiles.
We also acknowledge the limitations of our study. First, because of low rates of abdominal trauma in children under 1 year, this group of the youngest patients could not be examined. We acknowledge that reference values in this age group would be particularly helpful especially for infants with chronic liver disease such as biliary atresia, the most common indication for paediatric LT. While CT in children less than 1 year solely for tPMA determination is difficult to justify given the exposure to ionizing radiation, Tandon and colleagues recently showed excellent agreement between tPMA assessed using CT and magnetic resonance imaging, the latter being free of ionizing radiation. In the future, magnetic resonance imaging could therefore be considered in these youngest of children to measure tPMA. Second, we were unable to standardize tPMA by height because of the very limited height data available in this trauma data set. However, Golse et al. recently showed that tPMA alone performed identically to a tPMA:height2 index for predicting 1‐year post‐LT mortality in adults with area under the curve of 0.753. Finally, our population was on average 3–4 kg heavier than standard CDC curves. These CDC curves were derived in the early 2000s, whereas our population spanned 2005–2015, such that the differences are likely to reflect secular trends in body weight in North America.
In conclusion, we have defined novel paediatric reference curves for total psoas muscle area as a marker of sarcopenia in children, along with an online calculator to facilitate ease of use. These tools will greatly facilitate the identification of children at risk for sarcopenia to enable early implementation of targeted therapies to avoid associated comorbidity and optimize outcomes. In addition, these tools have important research applications and will help elucidate the relationship between sarcopenia and frailty and, ultimately, the prognostic significance of these parameters.
The authors of this manuscript certify that they comply with the ethical guidelines for authorship and publishing in the Journal of Cachexia, Sarcopenia and Muscle. We thank Brian Kirkham for helping with the creation of the tPMA calculator.
Eberhard Lurz has received speaker honoraria from Nutricia and Abbvie, but there is no conflict with this manuscript. Binita Kamath has received an unrestricted educational grant from Mirum. She is a consultant for Mirum, Shire, and DCI. None of these activities are in conflict with this manuscript. Peter Jüni serves as unpaid member of the steering group of trials funded by Astra Zeneca, Biotronik, Biosensors, St. Jude Medical, and The Medicines Company and has received research grants to the institution from Astra Zeneca, Biotronik, Biosensors International, Eli Lilly, and The Medicines Company and honoraria to the institution for participation in advisory boards from Amgen but has not received personal payments by any pharmaceutical company or device manufacturer. Vicky L. Ng is a consultant for Albireo, but there is no conflict with this manuscript. Hiten Patel, Gerald Lebovic, Claudia Quammie, Jessica P. Woolfson, Manuela Perez, Amanda Ricciuto, Paul W. Wales, and Govind B. Chavhan declare that they have no conflict of interest.
This work is supported by the SickKids Transplant and Regenerative Medicine Center and the Ashley's Angels fund.
© 2020. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
1 Division of Gastroenterology, Hepatology and Nutrition, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; Transplant and Regenerative Medicine Centre, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; Division of Gastroenterology, Hepatology and Nutrition, von Haunersches Kinderspital, LMU, Munich, Germany
2 Department of Diagnostic Imaging, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; Department of Radiology, Birmingham Women's and Children's NHS Foundation Trust, Birmingham, UK
3 Applied Health Research Centre, Li Ka Shing Knowledge Institute, St Michael's Hospital, Toronto, Ontario, Canada
4 Division of Gastroenterology, Hepatology and Nutrition, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
5 Division of Gastroenterology, Hepatology and Nutrition, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; Transplant and Regenerative Medicine Centre, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
6 Department of Diagnostic Imaging, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
7 Division of Gastroenterology, Hepatology and Nutrition, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada; Department of Surgery, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada
8 Applied Health Research Centre, Li Ka Shing Knowledge Institute, St Michael's Hospital, Toronto, Ontario, Canada; Department of Medicine, Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada