Comparison of Computed Tomography Data With Routine Measurements Concerning Bone and Muscle Health of Aged Individuals

Age-related muscle wasting and bone loss are significant public health challenges impacting elderly mobility and independence. Sarcopenia, a decline in muscle strength and mass, heightens fall risk, particularly in individuals with dementia or cognitive impairments. This leads to complications, reduced independence, and diminished quality of life. Osteoporosis increases the risk of fractures from falls or spontaneously, necessitating early diagnosis and prevention.

Despite effective treatments for sarcopenia, such as a protein-rich diet and strength training, the condition remains underrecognized due to diagnostic challenges. Common methods like hand strength measurement are problematic for those with rheumatic conditions or Parkinson's, often yielding inaccurate results. Other methods, like measuring leg strength and walking speed, require coordination and balance, which can be difficult for those with dementia or visual impairments. There is a need for tailored diagnostic solutions for diverse aging populations.

Bone mineral density (BMD) is crucial for assessing health in older adults, as low BMD indicates osteoporosis and a higher fracture risk. Traditionally measured by dual X-ray absorptiometry (DEXA), BMD assessment can be difficult for patients with mobility issues, pressure ulcers, or dementia due to the requirement to remain still for extended periods.

New AI-based algorithms can now automatically evaluate body tissues and patterns from routine CT scans, offering reproducible results beyond human capability. AI can quantify muscle mass at specific body cross-sections, such as lumbar vertebral point 3 (L3), which correlates with total body muscle mass and predicts muscle health. CT measurements of thigh and psoas muscles can also indicate whole body skeletal muscle mass. European guidelines highlight the need for muscle quantification in early sarcopenia diagnosis.

Correlating AI-measured muscle mass with functional muscle strength assessments can help identify surrogate parameters for early sarcopenia detection. Additionally, measuring muscle fat content, which correlates with strength loss, is essential for assessing muscle health. Innovative approaches are required, as sarcopenia diagnosis is still evolving and geriatric diseases often need proportionate diagnostic and treatment strategies due to multiple comorbidities.

For osteoporosis diagnostics, AI can determine Hounsfield Unit (HU) values from CT images to represent bone density, which can be correlated with DEXA results. These surrogate diagnostics should follow current guidelines, with results obtained within an appropriate interval of up to 18 months.

Based on the primary endpoints, the prevalence of "muscle-healthy" and "probably sarcopenic" individuals will be recorded. Additionally, the prevalence of individuals without osteoporosis, with osteopenia, and with confirmed osteoporosis will be calculated and included in the analysis.

The secondary endpoints compare functional muscle strength measurements with retrospective quantitative CT results. This exploratory analysis will test whether the AI algorithm can correlate functionality with muscle volume. Additionally, AI-measured muscle fatness (myosteatosis) could serve as a new, quantifiable quality criterion for muscle health. This is important because many elderly individuals cannot meet functional strength test requirements due to conditions like visual impairment, dementia, chronic pain, joint diseases, and frailty.

For osteoporosis diagnostics, the investigators aim to correlate bone density from DEXA measurements with CT-derived bone density of thoracic and abdominal vertebral bodies. This exploratory surrogate measurement method could provide insights into bone mineral density and health from existing CT images. CT-based diagnosis could be an alternative for those unable to remain still for DEXA scans.

Design:

This retrospective study will be conducted at a single center. Participants will be identified using databases of patients who have undergone CT and DEXA scans. To be included in the sarcopenia arm, participants must have had a CT scan from the radiology department at the University Hospital Basel within one month of an inpatient stay. For the osteoporosis arm, a thorax and/or abdomen CT scan and a DEXA scan must be available, conducted within 18 months of each other.

Anonymized DICOM datasets from the CT scans will be analyzed by an AI algorithm. Data analysis will involve standard statistical comparison using Student's t-test. The algorithm's values will be tested against reference standards, and its diagnostic accuracy will be evaluated for various diseases. The algorithm has been tested on 104 anatomical structures, organs, and organ groups (Req-2022-00495).

Origin of the data:

At the largest geriatric medical center in Switzerland, the investigators would like to include geriatric patients aged 65 and over who were undergoing inpatient treatment in the period from 01.07.2017 to 31.12.2022 inclusive.

By testing alternative diagnostic procedures (exploratory approach), the investigators want to reach more people, especially those who are unable to follow current diagnostic procedures due to dementia, visual impairment or post-operative condition. This is crucial as these patients in particular have a significantly increased risk of falling. The research approach aims to address real-world challenges in order to consider prevention programs and personalized therapies for as many people as possible in the future.

By using existing data and dispensing with further investigations, the economic aspect in the context of rising healthcare costs is also taken into account.

Scientific methodology and targets:

Exploratory analysis of quantitative variables that can be determined by CT to evaluate the primary endpoints:

• Sarcopenia: By evaluating muscle volume and fat percentage in muscle (in mm3) and density (in HU) on CT.
• Osteoporosis: By evaluating bone density/attenuation in HU on CT. For the evaluation of muscle volumes and densities, a feasibility study has already been carried out on a representative sample of over 4000 patients, which suggests a normal distribution of muscle density and volume (21). Statistical deviations from this reference group are determined using a t-test or Wilcoxon test in the case of a lack of normal distribution. The test for normal distribution is performed using the Kolmogorov-Smirnov test. The significance level is set at α = 0.05.

The evaluation of osteoporosis is performed analogously. The relationship between attenuation in HU and BMD has already been documented in the literature (22) (18).

Inclusion of as many patients as possible is essential for representative results

For which health-related personal data should consent be granted? A general consent form "Declaration of consent for the further use of health-related data and samples" was introduced at the beginning of 2020. Unfortunately, in practice it proved to be unreasonable to implement.

The investigators consider the sample size to be n=300. No consent has been obtained for any of the participants, meaning that an application for exemption is being made for all persons.

Quantitative, clinical data:

Hand strength measurement on both hands (geriatric routine assessment to assess the muscle strength of the upper extremity)

• Gait speed (geriatric routine assessment to assess mobility and muscle strength of the lower extremity; important for predicting falls)
• Timed up \& go (geriatric routine assessment to assess muscle strength of the lower extremity, coordination and sense of balance; important for predicting falls)
• Vitamin D level in serum (underestimated vitamin in old age, which is produced by the body itself through natural sunlight exposure of the skin and good kidney function in the investigators latitudes through many intermediate steps and contributes significantly to bone and muscle health; in the investigators latitudes (angle of incidence of the sun's rays), sufficient vitamin D production is only possible in the months of April to October with sufficient time spent outdoors. With age, however, the ability to synthesize vitamin D production also decreases considerably in the spring to summer months.

Quantitative data from imaging:

• Previous DEXA measurements of bone density (T-score, Z-score)
• Muscle mass and fat content in the muscle from previous CT scans (thorax, abdomen, pelvis, spine with muscle parts shown)