Patients who were deficient also had significantly more CD209+ DCs than those who were insufficient (P = 0·003). Furthermore, those who were VD3-insufficient or -deficient also had significantly higher circulating levels of CD1c+ DCs compared to healthy controls (P = 0·0003 and P < 0·0001, respectively). As shown in Fig. 3d, a strong inverse correlation exists between circulating

CD86+ DCs and VD3 status (R2 = 0·8501, P < 0·0001). VD3 also correlated inversely with PBMC expression RG7204 in vivo of CD209+ (Fig. 3e) (R2 = 0·7977, P < 0·0001), CD1c (Fig. 3f) (R2 = 0·8404, P < 0·0001) and CD1a (R2 = 0·9197, P < 0·0001, data not shown). Of the nine CRSwNP patients with CD209+ measurement, five had negative allergy testing, three had positive allergy testing and one was untested. Further evaluation determined that there were no significant differences between circulating CD209+ DCs levels in atopic versus non-atopic CRSwNP individuals (data not shown, P = 0·88). This would suggest that while atopic status may contribute to elevated numbers of DCs, such as in AFRS, there are mechanisms such as VD3 deficiency that result in an altered immune profile independent of atopy. While the Doxorubicin nmr CRSsNP cohort was overall VD3-sufficient, a correlation

analysis was conducted between VD3 and CD68+. As expected, there was no association between VD3 and circulating numbers of CD68+ cells (data not shown; R2 = 0·08, P = 0·72). Similarly, there was no correlation between VD3 plasma levels and circulation CD14+ monocyte levels among any of the cohorts (data not shown; R2 = 0·015, P = 0·71). Next we assessed plasma levels of macrophage and DC regulatory products, GM-CSF and PGE2. Figure 4a,b demonstrates that compared Amoxicillin to control, GM-CSF and PGE2 were increased in CRSsNP (P = 0·02 and P = 0·0011, respectively), CRSwNP (P < 0·0001 and P = 0·0004, respectively) and AFRS (P = 0·0067 and P = 0·0057, respectively). Levels of GM-CSF were also significantly higher in CRSwNP and AFRS compared to CRSsNP (P = 0·03 and P = 0·01, respectively) and levels of PGE2 were significantly higher

in AFRS compared to CRSsNP (P = 0·005). There was no statistically significant difference between CRSsNP and CRSwNP plasma PGE2 levels (P = 0·08). Similar to the DCs/VD3 correlation, VD3 correlated inversely with GM-CSF (R2 = 0·7039, P = 0·0012) (Fig. 4c) and PGE2 (Fig. 4d) (R2 = 0·7401, P = 0·0081). These results demonstrate that VD3 deficiency is associated with elevated levels of circulating DCs and DC regulatory products in CRSwNP and AFRS. VD3 has long been known as a regulator of bone health due to its ability to stimulate calcium absorption. Therefore we measured the severity of bone erosion on preoperative CT scans in patients with varying levels of VD3. As shown in Fig. 5a, the average CT bone remodelling score in patients with insufficient levels (<32 ng/ml) of serum VD3 was significantly greater than in patients with adequate (≥32 ng/ml) VD3 (P = 0·016) levels.