Background:​ Periostin (POSTN) regulates multiple biological behaviors of tumor cells. We studied the association of serum POSTN with mortality in home-dwelling men.

Methods:​ POSTN was measured at baseline using immunoassay (USCN life science, China) in 815 home-dwelling men aged 60–87 followed-up for 8 years.

Results:​ In the entire cohort, higher serum POSTN was associated with higher all-cause mortality [Hazard Ratio (HR) = 1.30 per Standard Deviation (SD), 95% Confidence Interval (CI):​ 1.13–1.50, p < .001] after adjustment for potential confounders. In a similar model, cancer mortality (n = 69) increased with POSTN levels (HR = 1.44 per SD, 95%CI:​ 1.16–1.78, p < .001). Cardiovascular mortality (n = 55) and non-cardiovascular-non-cancer mortality (n = 44) was not associated with POSTN levels (p = .28 and p = .94 respectively). In 107 men with prevalent cancer, all-cause mortality (HR = 1.93 per SD, 95%CI:​ 1.30–2.87, p < .005) and cancer mortality (HR = 2.07 per SD, 95%CI:​ 1.23–3.47, p < .01) increased with the increasing POSTN concentrations. In 613 men with incident cancer, higher POSTN concentrations were associated with higher cancer mortality (HR = 1.40 per SD, 95%CI:​ 1.12–1.76, p < .005) but not with the risk of cancer (HR = 1.16 per SD, 95%CI:​ 0.89–1.46, p = .21).

Conclusions:​ Higher serum POSTN is associated with higher cancer mortality, but not with the cancer risk in older home-dwelling men.