Computed tomography-based sarcopenia and pancreatic cancer: Prognostic value and variability based on different methods and thresholds

Abstract

Objective

To evaluate the prognostic value of Computed Tomography (CT)-determined sarcopenia on overall survival in patients with pancreatic cancer, and to assess the variability in outcomes according to different criteria described in the literature.

Materials and methods

This was a retrospective observational study including patients with histologically confirmed pancreatic adenocarcinoma treated at our institution. Two automatic artificial intelligence (Comp2Comp and TotalSegmentator) segmentation methods were applied to extract various body composition parameter values. Sarcopenia was defined according to the criteria proposed by Prado et al. (2008) (<52.4 cm²/m² in men and <38.5 cm²/m² in women), as well as using additional thresholds described in previous studies. Descriptive analyses (univariate and bivariate) and survival analyses were performed using Kaplan–Meier curves, Log-Rank tests, and univariate and multivariate Cox regression models.

Results

A total of 71 patients were included (39.4% female; mean age: 64.1 years), with a median overall survival of 10 months. The prevalence of sarcopenia according to Prado et al. was 59.2%, ranging from 5.6% to 70.4% depending on the alternative methods described in the literature. Sarcopenia defined by Prado et al. was associated with reduced overall survival (crude Hazard Ratio [cHR]: 2.03; 95% CI: 1.19–3.45). However, substantial variability was observed when using other criteria, with cHR values ranging from non-significant to as high as 6.05 (95% CI: 2.04–17.96).

Conclusions

CT-determined sarcopenia is a prognostic factor for overall survival in patients with pancreatic cancer. However, there is a need to harmonise and standardise the methods and thresholds used for its assessment.

Resumen

Objetivo

Analizar el valor de la sarcopenia basada en TC como factor pronóstico de la supervivencia global en pacientes con cáncer de páncreas y evaluar la variabilidad según diferentes criterios descritos en la literatura.

Material y métodos

Estudio retrospectivo observacional de pacientes con adenocarcinoma de páncreas atendidos en nuestro centro. Se aplicaron dos métodos automáticos de segmentación basados en inteligencia artificial (Comp2Comp y TotalSegmentator) para extraer los valores de diferentes parámetros de composición corporal. La sarcopenia fue definida según los criterios de Prado et al. (2008) (<52,4 cm²/m² en hombres y <38,5 cm²/m² en mujeres) y mediante otros umbrales utilizados en estudios previos. Se realizaron análisis descriptivos (univariantes y bivariantes) y de supervivencia mediante curvas Kaplan–Meier, pruebas Log-Rank y modelos de regresión de Cox univariantes y multivariantes.

Resultados

Se incluyeron un total de 71 pacientes (39,43% mujeres; edad media: 64,13 años) con una mediana de supervivencia de 10 meses. La prevalencia de sarcopenia según Prado et al. fue del 59,15%, variando entre el 5,63% y el 70,42% según otros métodos descritos en la literatura. La sarcopenia según Prado et al. se asoció a una menor supervivencia (Hazard Ratio cruda [HRc]: 2,03; IC95%: 1,19–3,45). Sin embargo, se encontró una variabilidad considerable al aplicar otros criterios, desde HRc no significativos hasta valores de 6,05 (IC95%: 2,04–17,96).

Conclusiones

La sarcopenia basada en TC es un factor pronóstico de la supervivencia en pacientes con cáncer de páncreas. Sin embargo, es necesario homogeneizar y estandarizar los métodos y umbrales utilizados para su determinación.