Background. Mortality related to human immunodeficiency virus (HIV) in
fection occurs predominantly in patients with CD4+ lymphocyte counts o
f less than 50 cells/mm(3). We followed 133 HIV-infected patients with
enrollment CD4 counts of less than 50 cells/mm(3) to determine if the
risk of death during a 1-year period could be predicted by a single e
nrollment CD8+ lymphocyte count. Methods. Enrollment data including ag
e, sex, T-cell subset counts, p24 antigen status, antiretroviral use,
and preexisting HIV-related illnesses were collected on a cohort of 13
3 consecutive patients with enrollment CD4 counts of less than 50 cell
s/mm(3). The cohort was followed for 1 year, and survival data were an
alyzed in relation to enrollment variables. Results. The mean enrollme
nt CD8 count of those patients alive at 1 pear was 600 cells/mm(3), co
mpared with a mean enrollment CD8 count of only 370 cells/mm(3) in pat
ients who had died prior to 1 pear (P <.001). For every 100-cell decli
ne in the enrollment CD8 count, the risk of death increased by 16% (95
% confidence interval [CI], 5% to 22%), independent of other enrollmen
t variables, including CD4 counts and p24 antigen status. A significan
t CD8 count warning level of 415 cells/mm(3), irrespective of the pres
ence of other enrollment variables, was associated with death within 1
year. The Kaplan-Meier estimated chance of death within 1 year was 54
% (95% CI, 42% to 66%) for patients with CD8 counts of less than 415 c
ells/mm(3) compared with only 25% (95% CI, 14% to 36%) for patients wi
th CD8 counts greater than 415 cells/mm(3). Conclusions. This study fi
nds that a single CD8 count has important prognostic significance in p
atients with advanced HIV infection and suggests that potential therap
ies to enhance CD8 counts might be beneficial to patients with advance
d HIV infection.