343   Prediction of CD4⁺Cell Response to Subcutaneous Recombinant Interleukin-2 (SC rIL-2).

N. Markowitz*¹, J. Bebchuk², E. Denning², and D. Abrams³.
¹Henry Ford Hosp., Detroit, MI;²Univ. of Minnesota, Minneapolis; and³Univ. of California, San Francisco.

Background:CPCRA 059 was a multicenter, open-label, randomized trial studying the virologic and immunologic impact of two doses of SC rIL-2 in HIV⁺patients (pts) with>300 CD4⁺cells/mm³on combination antiretroviral therapy (ART). We examined predictors of CD4⁺cell count response to SCrIL-2.

Methods:Pts were randomized to SC rIL-2 at 7.5 or 4.5 MIU b.i.d. (1:1). SC rIL-2 was administered for 5 consecutive days every 8 wks for at least 3 cycles. CD4⁺cell counts were measured at baseline (BL), every 4 mos, and 29 days after the start of each 5-day SC rIL-2 dosing cycle. Stratified analysis of covariance was used to compare CD4⁺change from BL to day 29 of cycle 3 and to month 12. For estimated regression coefficients (b), CD4⁺and body mass index (BMI) were modeled as continuous variables.

Results:Of 256 participants randomized to receive SC rIL-2, 126 and 130 pts were assigned the 7.5 and 4.5 MIU doses, respectively: 192 pts completed>3 SC rIL-2 cycles and 245 had a month 12 CD4⁺cell count. At BL viral load was <50 copies/ml in 60% of pts. Median pre-enrollment nadir and BL CD4⁺cell counts were 293 cells/ mm³and 538 cells/mm³

respectively. Pts received prior ART for a median of 42.2 mos. The strongest predictor of CD4⁺change from BL to month 12 was nadir CD4⁺(b= 0.34, p < 0.0001). Among those who completed>3 cycles of SC rIL-2, nadir CD4⁺(b= 0.81, p < 0.0001) and BMI (b= -21.5, p < 0.004) were the strongest predictors of CD4⁺change from BL at day 29 after the third cycle. BL CD4⁺had no significant effect (b= 0.09, p < 0.62).

Conclusions:Nadir CD4⁺was the strongest predictor of SC rIL-2 response, emphasizing the persistence of immune defects despite normalization of CD4⁺cell count with ART. These data suggest that BMI should be considered in selecting SC rIL-2 dosing.