Of 461 participants, 360 (78.1%) had available biosamples for the current study. Baseline characteristics of subjects were similar to the overall DELIGHT population and were balanced between treatment allocation (Table 1). Correlation coefficients between inflammation markers and iron parameters and erythropoietin at baseline are shown in Additional file 1: Table S1. Serum IL-6 was negatively correlated with iron concentration (r= -0.18, p < 0.001) and TSAT (r= -0.14, p = 0.008). Urinary IL-6/Cr and erythropoietin did not correlate with any markers assessed.

Changes in biomarkers by treatment allocation and their differences relative to placebo are presented in Fig. 1. In the placebo group, hemoglobin decreased during the study period (mean change − 1.5 g/L [-2.5, -0.6], p < 0.001). Compared to placebo, dapagliflozin and dapagliflozin-saxagliptin increased hemoglobin by 5.7 g/L (4.0, 7.3), p < 0.001, and 4.4 g/L (2.7, 6.0), p < 0.001, respectively. There was no clear difference in hemoglobin change between dapagliflozin and dapagliflozin-saxagliptin (-1.3 g/L [-2.9, 0.34], p = 0.16) Similarly, both treatments also increased hematocrit compared to placebo.

Changes in markers of hematopoiesis, iron metabolism and inflammation. Mean absolute changes in (A) hemoglobin and (B) hematocrit over time. Mean percentage changes in (C) serum iron, (D) transferrin, (E) TSAT, (F) ferritin, (G) erythropoietin, (H) urinary MCP-1/Cr, (J) urinary IL-6/Cr and (K) serum IL-6. Numbers across bars represent between group differences in change relative to placebo group. Cr, creatinine; MCP-1, monocyte chemoattractant protein; IL-6, interleukin-6; TSAT, transferrin saturation

There was a significant reduction in ferritin in both dapagliflozin alone and dapagliflozin-saxagliptin group compared to placebo (-18.6% [-27.5, -8.7], p < 0.001 and − 18.4% [-27.1, -8.7], p < 0.001, respectively) with no difference in iron concentration (Fig. 1). Transferrin increased in the dapagliflozin group (4.0% [1.3, 6.8], p = 0.004) compared to placebo, but not in the dapagliflozin-saxagliptin group (1.2% [-1.4, 3.9], p = 0.36). Accordingly, the increase in transferrin in dapagliflozin was higher than that in dapagliflozin-saxagliptin (2.7% [0.1,5.2], p = 0.04) There were numerical trends to decrease TSAT and increase EPO among patients treated with dapagliflozin, although these were not statistically significant. There was no clear difference between dapagliflozin and dapagliflozin saxagliptin on iron, TSAT, ferritin and EPO (all p values > = 0.15).

Dapagliflozin reduced urinary inflammatory markers (Fig. 1). Between-group differences in urinary MCP-1/Cr change versus placebo were − 29.0% (-41.0, -14.6), p < 0.001, and − 29.0% (-40.7, -14.9), p < 0.001 for dapagliflozin alone and the dapagliflozin-saxagliptin group, respectively. Changes in urinary IL-6/Cr with dapagliflozin and dapagliflozin-saxagliptin were − 26.6% (-40.7, -9.1), p = 0.005, and − 32.3% (-45.1, -16.6), p < 0.001, respectively. Serum IL-6 did not change in each treatment group. There was no difference in the effect on MCP-1/Cr, IL-6/Cr and serum IL-6 between dapagliflozin and dapagliflozin-saxagliptin (all p values > = 0.44).

Correlation coefficients between changes in makers from baseline to week 24 by treatment were shown in Table 2. Negative correlations between changes in serum IL-6 and iron and TSAT were observed in all treatment groups (all p ≤ 0.01). In the dapagliflozin group, decreases in urinary MCP-1/Cr and IL-6/Cr were correlated with increases in iron and transferrin. Urinary IL-6/Cr change was correlated with changes in TSAT and erythropoietin only in the dapagliflozin group.