baseline viral load, history of opportunistic infection, co-morbidities, and age. Potential mechanisms underlying the low-level regeneration of CD4 cells in immune non-responders include deficiencies in the regeneration of central memory CD4 cells and excessive apoptosis8. Increased CD4 and CD8 T cell activation at baseline has also been correlated with poor immune responses to HAART9. In one early observational study, no association was found between a specific antiretroviral regimen and the presence of discordant responses5. In recent years, however, a number of reports have described CD4 declines or attenuated gains despite viral suppression in patients receiving a combination of didanosine and tenofovir10. While co-administration of these two agents has been shown to increase the ddI serum concentration, thereby increasing the risk of ddI-associated toxicities, the exact mechanisms leading to the CD4 effects are uncertain. The effect is diminished when a lower ddI dose is used11. Failure to increase CD4 counts may also be associated with additive myelotoxicity of antiretroviral drugs and therapies being administered for the prevention or treatment of opportunistic infection, as, for example, the additive myelosuppression of zidovudine (AZT) and trimethoprim/sulfamethoxazole (TMP/SMX). It has been postulated that ongoing, low-level viral replication, occurring below the limits of assay detection, may contribute to slowed CD4 T-cell recovery. Consequently, one approach has been to change or intensify antiretroviral regimens in an effort to more fully suppress viral replication and allow immune recovery, though available data do not support this practice. Alternative strategies aimed at improving the HIV-induced immunodeficiency utilize immune-based therapies. Potential approaches that have been studied over the years include therapy with cytokines such as interleukin-2 (IL-2), IL-12, IL-7, interpheron alpha, interpheron gamma and GM-CSF, as well as cell-transfer approaches using unmodified or genetically modified CD4 or CD8 cells. Immunosuppressive medications aiming to reduce high level immune activation thought to underlie CD4 cell turnover and death have also been studied, including cyclosporine A, anti-TNF monoclonal antibodies (etanercept), and mycophenolate mofetil. The best-studied and most promising immunotherapy agent for HIV infection that has been investigated to date is IL-2. IL-2 is a cytokine produced by T lymphocytes that is known to promote proliferation and modulate the secretory capacity of lymphocytes, including T, B, and natural killer cells. Numerous randomized trials have demonstrated that intermittent subcutaneous or intravenous administration of cycles of IL-2, in combination with antiHighly active antiretroviral therapy (HAART) has dramatically reduced the morbidity and mortality associated with HIV infection. In most patients receiving HAART, suppression of HIV replication is accompanied by recovery of CD4 T lymphocytes to near normal levels and a substantial reversal of HIV-associated immunologic defects, as is best reflected by the marked decrease in HIV-related opportunistic complications. However, immunologic recovery varies greatly between patients, and a significant proportion of those treated experience only small improvements in CD4 cell counts despite virologic suppression on HAART. These patients, who maintain virologic suppression on HAART but fail to have significant gains in CD4 T-cell counts, are referred to as immunologic non-responders. Estimates of the frequency of immunologic non-response vary depending on the definition used. Based on a number of cohort studies, 10% to 20% of treated patients fail to have an adequate CD4 count increase (e.g > 100 cells/mm3 increase over baseline or an increase to > 200-300 cells/mm3) after 6-12 months of effective HAART1-7. Whereas some patients with a poor initial immune response despite virologic suppression will go on to have an immunologic response over time, others will be stalled at a threshold that potentially puts them at increased risk for opportunistic complications. Observational cohort studies suggest that such patients do in fact have a higher rate of new AIDSdefining events or death than complete responders, though these rates are less than is seen in patients who are both immunologic and virologic non-responders2,6,7. The pathogenesis of discordant responses is poorly understood. The immunologic response to HAART is influenced by a number of viral, host, and treatment-related factors. Immune recovery depends, in part, on the extent and duration of viral load reduction, although the minimum viral suppression necessary for improved immune function and CD4 recovery is debated. Numerous other factors may influence CD4 T cell and immune recovery, including the degree of immunosuppression at HAART initiation (e.g., baseline CD4 count), duration of infection,