br Regression models for cachexia classification stages
Regression models for cachexia classification stages according to outcomes.
PG-SGA SF (global score)
Symptoms of nutritional impact
Note: OR ¼ odds ratio; CI ¼ confident interval; HR ¼ hazard ratio; BMI ¼ body mass index; MUAMA ¼ mid-upper arm muscle area; PG-SGA SF¼ Patient-Generated Subjective Global Assessment Short Form; NLR ¼ neutrophil/lymphocyte ratio; PLR ¼ platelet/lymphocyte ratio; WL ¼ weight loss; HGS ¼ hand grip strength.
a Logistic and Cox regression adjusted for age 60 years, female gender, type of tumor, Karnofsky Performance Status 30 and 40%; current medical situation e in patient versus out patient.
c According PG-SGA SF.
e According Edmonton Symptom Assessment System.
Additionally, we observed a lower prevalence of patients in PCa group (3.9%), which may reduce the statistical test accuracy.
Other studies failed to differentiate the PCa stage from the other stages [3,27]. Vigano et al.  proposed a cancer cachexia classifi-cation based on clinical features, however they were unsuccessful in distinguishing the pre-cachectic from the cachectic group. They justified that in the PCa group, there is a possibility of coexistence of patients with high cachexia risk and early stages of the syndrome. Thereby, we supposed, based on the dynamic nature of cancer cachexia that only cross-sectional observations of the inflammatory biomarkers should be insufficient to define PCa.
Fig. 2. KaplaneMeier survival curves stratified according to cachexia classification stages. Note: aStatistically different from No cachexia; bStatistically different from Undernourished; cStatistically different from.
As would be expected, significant differences in the risk of death at 90-days were observed for all the cachexia stages. The cachexia classification system based on systemic AMG-176 criteria showed to be a better survival predictor when compared with another classification system based on clinical features [3,27]. As already discussed, Gray and Axelsson  reported a progressive increase in the prevalence of CRP >10 mg/dL and albumin <30 g/L when closer to death. In a study with inoperable pancreatic cancer patients, Bye et al.  showed that albumin decreased significantly from 43 mg/mL to 39 mg/mL (p ¼ 0.01), whereas CRP increased from 5.8 mg/mL to 14.1 mg/mL. The median survival of this group was of 5.5 months.
The biggest limitation of this study was the evaluation of patients in a cross-sectional manner. Despite the fact that CRP was used as the most relevant biomarker for cachexia inflam-mation haploid is not specific for cancer, cachexia or for tumor ac-tivity, since it can be influenced by other factors such as infections. Moreover, although the mGPS framework clearly distinguishes NCa and RCa stages for all domains analyzed, it was unable to capture all stages of cachexia. Due to the lack of statistical discrimination between the PCa stage and almost all the outcomes examined, there is a need for further exploration, aiming a validation of this method with other clinical charac-teristics focusing on the benefits for cachectic advanced cancer patients.
Cachexia stages evaluated by mGPS were associated with poor clinical features and can predict OS. This classification system based on simple and objective criteria available in routine clinical practice can be used to identify and characterize the presence and severity of cachexia in advanced cancer patients. 97
Please cite this article as: Alves da Silva G et al., Clinical utility of the modified Glasgow Prognostic Score to classify cachexia in patients with advanced cancer in palliative care, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.07.002
6 G. Alves da Silva et al. / Clinical Nutrition xxx (xxxx) xxx
1 Conflict of interest
3 None declared.