Race-based differences in serum biomarkers for cancer-associated cachexia in a diverse cohort of patients with pancreatic ductal adenocarcinoma

Abstract

Background: Pancreatic ductal adenocarcinoma is projected to become the second leading cause of cancer-related deaths by 2040, with the highest disease burden expected amongst Non-Hispanic Black patients. One of the most significant predictors of poor outcomes is the presence of cancer-associated cachexia (CCa). Yet, race- and ethnicity-specific biomarkers for early CCa diagnosis are lacking.

Methods: We evaluated a panel of candidate biomarkers of CCa in a diverse cohort of patients with pre-treatment serum using multiplex ELISA-based methods.

Results: We find that growth/differentiation factor-15 (GDF-15) is associated with cachexia severity, is superior to standard biomarkers at classifying cachexia, and differentiates between non-cachexia and pre-cachexia status, but only among Hispanic/Latinx and non-Hispanic White participants. Furthermore, high GDF-15 levels at diagnosis are associated with a greater weight loss from 3.3% (95%CI = -0.14-6.7) to 8.0% (CI = 5.9-10.1) over the 6 months post-diagnosis. Finally, both ENA-78/CXCL5 and GRO-α/CXCL1 are elevated in non-Hispanic Black individuals in a disease-independent manner (P < 0.001 for both analytes).

Conclusions: GDF-15 may be a potential biomarker for "pre-cachexia" in the non-Hispanic White and the Hispanic population, but not non-Hispanic Black individuals. These findings underscore the unmet need to enroll non-Hispanic Black participants in clinical trials for CCa.

Fig. 1. Cachectic and pre-cachectic PDAC patients demonstrate significant differences in stress-related markers at baseline and GDF-15 can be used to classify patients into cachexia categories.

Boxplots showing a log-scale value of all significantly different serum biomarkers between cachectic and non-cachectic PDAC patients (A) and when stratified into 4 stages as in Vigano et al. B NCa = non cachectic, PCa = pre cachectic, Ca = cachectic, RCa = refractory cachectic. All plots in (A, B) are significant based on a Wilcoxon rank sum or Kruskal-Wallis test and BH-adjusted p-value (two-sided hypothesis). Significant differences in serum biomarkers between groups (BH-adjusted pairwise Dunn’s test, two-sided hypothesis) are denoted with a red line in (B). Outliers are denoted by a red dot. C ROC curves for percent self-reported weight loss (Pct Weightloss), GDF-15, TNF-alpha, white blood cell count (WBC), Hemoglobin and Albumin and a combined TNF-alpha * GDF-15 model (Combined Model) using cachexia status at baseline as the “ground truth”. Youden’s optimal thresholds are denoted with a black dot. D Boxplot of percent weight loss (negative values indicate weight gain) from baseline to 6-month follow-up time point. Patients were dichotomized based on levels of GDF-15 (Low vs High based on the Youden’s threshold values calculated by the ROC in Fig. 2C or by the values used in Groarke et al. (37). P-values represent a Wilcoxon rank sum test (two-sided hypothesis).

Fig. 2. Both NHW and H/L participants demonstrate significant differences between non cachexia and cachexia status in stress-related markers at baseline.

Boxplots showing a log-scale value of serum biomarkers which are significantly different between cachectic and non-cachectic PDAC patients for at least two R/Es stratified into Non-Hispanic Black, Hispanic/Latinx and Non-Hispanic White patients. Significance was determined by Kruskal-Wallis test (BH-adjusted) followed by a BH-adjusted Dunn’s pairwise test (two-sided hypothesis). NHB Non-Hispanic Black, H/L Hispanic/Latinx, NHW Non-Hispanic White, NCa non cachectic, Ca cachectic. Significant differences in serum biomarkers between groups are denoted with a red line. Outliers are denoted by a red dot.

Fig. 3. PDAC and other patients demonstrate wide racial variation in select serum biomarkers at baseline.

A Boxplots showing a log-scale value of select significant biomarkers when PDAC R&E groups are compared (ENA-78/CXCL5 and GRO-α/CXCL1) and B when stratified into cachexia groups (NHB Non-Hispanic Black, HL  Hispanic/Latino, NHW Non-Hispanic White, NCa non cachexia, Ca Cachexia). C Boxplots showing ENA-78 and GRO-alpha levels for non-PDAC diagnoses. D Boxplot of neutrophil to lymphocyte ratio (NLR) for PDAC patients in the study. Significant differences between relevant groups are denoted with a red line. Significance between NHBCa/NHBNCa and all other groups is denoted with a *. Significance was determined by Kruskal-Wallis test (BH-adjusted) followed by a BH-adjusted Dunn’s pairwise test (two-sided hypothesis). N/S not significant. Outliers are denoted by a red dot.

Fig. 4. GDF-15 and IL-6 are linked to overall survival in both CoxPH models and by Kaplan estimation.

A Forest plot of dichotomized analytes found to be associated with survival as continuous variables in a univariate analysis when controlled for demographic and patho-clinical variables. NHB Non-Hispanic Black, H/L Hispanic/Latinx, NHW Non-Hispanic White, NCa non cachectic, Ca Cachectic. P-values represent a pairwise comparison of the adjusted CoxPH model relative to the reference group (two-sided hypothesis). Median levels of GDF-15 (B) or IL-6 (C) were calculated and participants were classed into either low (Low) or high (High) groups based on a median split. B, C represent Kaplan-Meier survival curves (Time = time to event in months) for GDF-15 (P = 0.003) and IL-6 (P < 0.0001), respectively. P-values represent global p-values of the survival analysis (two-sided hypothesis).

Fig. 5

Overview of IL-6 and GDF-15 levels throughout the cancer-associated cachexia continuum.