Progressive multifocal leukoencephalopathy (PML) is a fatal demyelinating disease of the CNS of immunosuppressed individuals caused by the polyomavirus JC (JCV). In previous studies, we showed that JCV-specific cytotoxic T lymphocytes (JCV-specific CTL) were associated with a favourable outcome in patients with PML. However, these CTL had been assessed in PML survivors more than 1 year after the onset of disease and we could not determine whether this immune response was only a surrogate marker for a general recovery of the patient's immune system or a causal factor in the patient's neurological improvement. In this study, we assessed the relationship between JCV-specific CTL detected early in the course of PML and the subsequent course of disease activity. We enrolled 26 patients with possible or proven PML, including 21 HIV+ patients, less than 10 months after the onset of their neurological symptoms (3.7 +/- 2.5 months, median +/- interquartile range). JCV-specific CTL were detected by either 51Cr release or tetramer staining assay. Patients were then followed prospectively and the clinical course of PML was determined. At the time of their first immune evaluation, we found that 15 patients had detectable JCV-specific CTL. HIV+ patients with JCV-specific CTL had a higher CD4+ T-cell count (215 +/- 103/microl) and a lower HIV viral load (144 +/- 431 copies/ml) than those without JCV-specific CTL (32 +/- 59/microl, P = 0.004 and 43 100 +/- 54 778 copies/ml, P = 0.01). Thirteen of these 15 patients with JCV-specific CTL developed clinically quiescent PML, while only two out of 11 without detectable CTL controlled their neurological disease. Therefore, the early detection of JCV-specific CTL had an 87% predictive value for subsequent control of PML, while the absence of such CTL had an 82% predictive value for subsequent active PML (P = 0.0009). Fifteen patients were evaluated less than 4 months after the onset of PML (1.9 +/- 1.3 months). Of nine patients with JCV-specific CTL, seven (78%) demonstrated subsequent control of disease, whereas six out of six (100%) without JCV-specific CTL developed progressive PML (P = 0.007). Two to ten CTL assays were performed on PBMC of 11 patients. Of these patients, one had an increase in JCV-specific CTL preceding a significant clinical improvement. In another patient with otherwise stable immune parameters, a decline in JCV-specific CTL preceded an exacerbation of PML. We conclude that JCV-specific CTL can be detected early in PML and can predict control of this disease. Fluctuations of JCV-specific CTL in the blood are associated with variation in disease manifestations. These results indicate that JCV-specific CTL are associated with the control of PML.