CT Body Composition Misses Cancer Nutritional Risk

Key Takeaways

• Among hospitalized adults with cancer at a tertiary center in São Paulo, Brazil, CT identified low skeletal muscle mass in 51.1% of patients and low skeletal muscle density in 52.2%, including some with normal or elevated BMI.
• For CT-defined low muscle mass, the institutional CNS was more sensitive but less specific than NRS-2002.
• Many CT-detected abnormalities were not captured by routine nutritional-risk classification.
• After adjustment, neither screening tool remained independently associated with low muscle mass or low muscle quality, while older age remained independently associated with low muscle quality, and the authors suggested CT assessment may complement screening when scans are already available.

Hidden muscle depletion and reduced muscle quality can be hard to recognize in hospitalized adults with cancer when bedside nutrition screening and body mass index do not fully reflect body composition. Abdominal imaging is often already available in inpatient oncology, creating a practical opportunity to compare routine screening with imaging-based muscle measures in São Paulo, Brazil. Hospitalized patients at A.C. Camargo Cancer Center had bedside nutritional screening compared with CT-based body-composition measures to test how closely those approaches aligned.

In a retrospective, cross-sectional study, Lima and colleagues in Supportive Care in Cancer included 272 hospitalized adults with confirmed cancer at A.C. Camargo Cancer Center in São Paulo, Brazil, from 2022-2023 after eligibility review. Abdominal computed tomography (CT) had to be available within 30 days before admission, and patients also needed documented Nutritional Risk Screening-2002 (NRS-2002) or the institutional Nutritional Screening Tool (CNS) during hospitalization; intensive care unit and exclusive palliative care admissions were excluded, and only the first hospitalization per patient was analyzed. CT measurements of skeletal muscle index (SMI) and skeletal muscle density (SMD) were taken from a single axial slice at the inferior border of L3 using CoreSlicer, with multivariable logistic regression adjusted for age, body mass index (BMI) category, and metastasis to compare screening status with CT-defined abnormalities.

CT-defined low muscle mass and low muscle density were common, including among patients whose BMI was normal or elevated, and routine nutritional-risk categories missed many of those abnormalities. The authors reported that many patients with CT-defined low muscle mass were not classified as nutritionally at risk by screening, although the paper’s tool-specific sensitivities were 76% for CNS and 52% for NRS-2002. For CT-defined low muscle mass, CNS showed 76% sensitivity and 30% specificity, a tradeoff that favored sensitivity over specificity.

NRS-2002 showed 52% sensitivity and 63% specificity for CT-defined low muscle mass and was the only screening tool with a significant bivariate association with SMI. In the primary adjusted models, however, neither NRS-2002 nor CNS was independently associated with low muscle mass or low muscle quality. Older age was independently associated with low muscle quality, with an odds ratio of 1.081 and 95% confidence interval of 1.055-1.108, while a sensitivity analysis excluding BMI left only a borderline NRS-2002 signal for low muscle mass.

Interpretation stays bounded by a retrospective, cross-sectional analysis from a single tertiary center in Brazil. Possible selection bias could have arisen because intensive care unit and exclusive palliative care patients were excluded, and the investigators could not systematically compare included and excluded patients on key clinical variables. Formal sarcopenia also could not be established because muscle strength and physical performance were unavailable, so the main implication is a detection gap in routine inpatient oncology nutrition workflows within this Brazilian setting rather than a broader claim about outcomes.

Within this hospital setting, commonly used nutritional screening tools showed limited agreement with CT-defined abnormalities in muscle mass and muscle quality. The authors framed opportunistic CT-defined body composition analysis as a complementary approach when abdominal imaging is already available, not as a replacement for screening and not as proven outcome-improving practice.

Clinician Questions

Which hospitalized cancer patients were included in the CT and nutritional screening comparison?

Hospitalized adults aged 18 years or older with confirmed cancer at a tertiary cancer center in São Paulo, Brazil, were eligible if they had an abdominal CT within 30 days before admission and documented nutritional screening during hospitalization. Only the first admission per patient was analyzed, and patients admitted to the intensive care unit or receiving exclusive palliative care were excluded.

How were low muscle mass and low muscle quality defined on CT in hospitalized adults with cancer?

Low muscle mass and low muscle quality were assessed on a single axial CT slice at the inferior border of L3 using CoreSlicer, with skeletal muscle area segmented from the psoas, paravertebral, and abdominal wall muscles. SMI was normalized to height, with depletion defined as less than 55 cm2/m2 in men and less than 39 cm2/m2 in women, and SMD was assessed with sex-specific cutoffs.

Why did the NRS-2002 signal for low muscle mass weaken after BMI adjustment?

The authors suggested that BMI is built into NRS-2002, so adding BMI to the multivariable model may have attenuated the association between NRS-2002 risk status and low SMI. They also noted that the sensitivity analysis excluding BMI left only a borderline association for low SMI.