The premorbid level of selenoprotein P in plasma from subjects with cancer
at different sites was compared with that from control subjects in a nested
case-control study. A health screening of 12,500 middle-aged men was perfo
rmed during 1974-1982 in Malmo, Sweden, and from the 400 cancer cases that
were identified during follow-up until the end of 1988, 302 plasma samples
were available for analysis of selenoprotein P. Two living controls per cas
e of the same screening day and age were chosen. Selenoprotein P levels in
subgroups of major cancer sites were lower in cases than in controls for th
e respiratory tract (1.20 and 1.30 arbitrary units, respectively; p < 0.05)
cancer group. The odds ratio for overall cancer risk in the lowest quintil
e of selenoprotein P level compared with that in the highest was 5.2 [p (fo
r trend) = 0.01]. In subgroups of major cancer sites, the odds ratios for c
ancer risk in the lowest tertile compared with the highest were 6.0 [p (for
trend) = 0.004] in the respiratory tract and 3.4 [p (for trend) = 0.002] i
n the digestive tract. In cases + controls, selenoprotein P was lower in sm
okers than in nonsmokers (9 < 0.05). Selenoprotein P was significantly corr
elated to plasma albumin, fasting blood glucose, and body mass index and in
versely correlated to plasma alpha(1)-antitrypsin and gamma-glutamyl transf
erase. The results suggest that a low plasma selenoprotein P level is assoc
iated with higher future risk of respiratory and digestive tract cancer in
middle-aged men.