Progressive multifocal leukoencephalopathy in patients treated with chimeric antigen receptor T cells

TO THE EDITOR:

Progressive multifocal leukoencephalopathy (PML) is a rare, devastating, and fatal demyelinating disease of the central nervous system (CNS) caused by the JC virus (JCV). It typically affects patients with profound immunosuppressive states secondary to AIDS or iatrogenic exposures.¹ CD19-directed chimeric antigen receptor T-cell (CAR-T) therapy is an effective treatment for B-cell malignancies.^2,3 However, complications such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are common. Furthermore, CAR-T is associated with immunosuppression secondary to off-target effects of CAR-T, lymphodepleting agents, cytopenia, and hypogammaglobulinemia.^4-6 We hypothesized that CD19-CAR-T therapy increases the risk of PML. Descriptions of post−CD19-CAR-T PML cases have been reported in anecdotal case reports and in a case series from the US Food and Drug Administration’s FDA Adverse Event Reporting System (FAERS).^7-10 However, the incidence of PML after CD19-CAR-T and whether CD19-CAR-T is associated with PML risk remain to be determined. We address these questions in this retrospective pharmacovigilance study using 2 global postmarketing surveillance databases.

To estimate increased reporting of PML following commercial CD19-CAR-T therapy, we compared the reporting of CD19-CAR-T−related PML with that of other drugs in the entire FAERS database. Patient-level data were obtained via a Freedom of Information Act request. To identify a significant disproportionality signal, we used the reporting odds ratio (ROR) and the lower bound of the information component 95% credibility interval (IC₀₂₅).^11,12 To increase signal detection specificity, we restricted the analysis to reports in which CD19-CAR-T was the primary suspect.

To estimate the incidence of PML, we queried the Center for International Blood and Marrow Transplant Research (CIBMTR) registry, which captures approximately 65% of all commercial CAR-T therapy performed in the United States (see supplemental Methods for additional details, available on the Blood website).¹³ 

Of 4,198,876 eligible patients in the full FAERS database, 4037 patients received CD19-CAR-T (axicabtagene ciloleucel, n = 2338; tisagenlecleucel, n = 1699) between 1 January 2018 and 31 March 2021. We identified 11 reports of PML in which CD19-CAR-T therapy (axicabtagene ciloleucel, n = 7; tisagenlecleucel, n = 4) was considered the primary suspect (Figure 1A). PML cases were reported mostly by healthcare professionals (n = 10 of 11 [91%]) and the events occurred in non-Hodgkin lymphoma (NHL) patients (n = 9 of 10 [90%]), apart from 1 patient with acute lymphoblastic leukemia (ALL) (Table 1; supplemental Table 1). The median age of PML patients was 63 (IQR = 58-67) years, slightly higher than for other CD19-CAR-T−treated patients reported to FAERS (57 [IQR = 28-67], P = .163). Female sex was enriched in the PML cohort compared to other FAERS patients (70.0% vs 39%, P = .052). A history of HIV or of natalizumab use was not reported in the patients with available data. However, other PML risk factors, including chemotherapy and rituximab exposure, were common. At the time of PML diagnosis, 5 patients had active hematological disease, and 4 patients were in complete remission after CAR-T. The CAR-T recipients who developed PML presented with varied neurological symptoms, most commonly motor and cognitive disorders (supplemental Table 1). Time to PML onset was variable, ranging between 6 and 31 months from CD19-CAR-T administration, with a median of 8 months (IQR = 8-15 months). In 9 patients with available data, PML diagnosis was confirmed in 5 and 3 patients by JCV presence in the cerebrospinal fluid and a brain biopsy, respectively. Notably, the reported case fatality rate of PML was 78% (n = 7 of 9) at last follow-up. PML treatment was described in 2 patients, 1 patient with pembrolizumab and the other with mefloquine, IVIG, and mirtazapine; both were alive at last follow-up.

To determine whether CD19-CAR-T therapy is associated with increased PML reporting, we performed a disproportionality analysis. PML reporting was higher in patients treated with CD19-CAR-T compared to unexposed patients (ie, remaining FAERS database) (ROR = 7.66 [95% CI, 4.23-13.88], IC₀₂₅ = 1.59). We further conducted several sensitivity analyses. First, in a stratified analysis by CAR-T product, both axicabtagene ciloleucel and tisagenlecleucel were significantly associated with increased PML reporting (n = 7, ROR = 8.42 [95% CI, 4.00-17.72], and n = 4, ROR= 6.62 [95% CI, 2.48-17.68], respectively; both IC₀₂₅ >0) (Figure 1B). Second, restricting the analysis to patients receiving CAR-T for NHL yielded consistent results (n = 9, ROR = 10.19 [95% CI, 5.29-19.65], IC₀₂₅ = 1.68). Finally, when restricted comparator groups were selected, increased reporting of CAR-T−related PML remained significant compared to that in cancer patients (ROR = 2.45 [95% CI, 1.35-4.45], IC₀₂₅ = 0.22), whereas compared to hemato-oncological patients, a numerically, non−statistically significant, higher ROR was observed (ROR = 1.47 [95% CI, 0.81-2.67], IC₀₂₅ = −0.45) (Figure 1B). Overall, these findings support an association between CD19-CAR-T therapy and greater PML risk.

The CIBMTR registry included 3239 patients who received commercial CD19-CAR-T cells between 1 January 2018 and 31 Decemeber 2020. We identified 2 CD19-CAR-T−related PML cases in NHL patients, reported 7.5 and 16 months after axicabtagene ciloleucel therapy (supplemental Table 2). At last follow-up, 1 patient was alive and the other deceased. Another case of a 12-year-old ALL patient presenting with neurological symptoms and radiological findings of leukoencephalopathy 9 months after tisagenlecleucel was reported. Although PML is most likely, a definitive diagnosis was not reported to the registry. Considering all 3 reported cases, the estimated incidence of PML following CD19-CAR-T is 0.9 (95% CI = 0.2-2.7) cases per 1000 CD19-CAR-T−treated patients.

The use of CD19-CAR-T therapy in hematological malignancies is rapidly increasing, with a growing number of indications and introduction at earlier stages of the disease. Thus, characterization of the full spectrum of CAR-T toxicity is a clinical priority. This post-marketing surveillance study suggests PML as a potentially rare and fatal complication of CD19-CAR-T therapy. The increased reporting of PML following CD19-CAR-T treatment was consistent across different statistical methods (ROR and IC₀₂₅) and multiple sensitivity analyses. PML case patients had a high fatality proportion (78%), consistent with previous studies that observed 52% to 83% mortality rates depending on the underlying cause.¹⁴ The highly variable time of PML onset is in line with previous studies of natalizumab and rituximab.^15,16 It may reflect variations in the host immune response and a pre-symptomatic stage, as well as potentially delayed diagnosis.¹⁶ Using the CIBMTR registry, we deduce an incidence of 0.9 cases per 1000 CAR-T recipients, similar to the incidence of PML in patients treated with natalizumab (1 per 1000), which is 1 of the most notable drug-PML associations.^16,17 As early PML diagnosis is associated with favorable outcomes,¹⁸ and as novel immune-based strategies for PML treatment have early promising results (eg, immune checkpoint blockade, recombinant human interleukin-7, and BK virus−specific T lymphocytes, which cross-react with JCV),^1,19-21 clinicians should be conscious of this potential complication.

PML develops almost exclusively in immunocompromised patients due to the reactivation of a dormant JCV.¹ Lymphopenia, neutropenia, B-cell aplasia, lymphodepletion regimens, and CRS treatment with interleukin-6 inhibitors or corticosteroids may all contribute to the immunosuppressive state in CAR-T recipients.^4,5 Cellular immunity is crucial in protecting against JVC reactivation, and PML is most common in patients with CD4+ T-cell deficiencies, as in HIV infection. Thus, prolonged impairment of cellular immunity in CAR-T recipients,^4,5 which may be precipitated by CAR-associated B-cell depletion,²² provides a biological rationale for an increased PML risk.^1,16 

This study has several limitations. Causality between CAR-T and PML occurrence cannot be established due to the inherent nature of the FAERS. Causes for JCV reactivation and PML are likely multifactorial and reflect the cumulative exposure to immunosuppressive interventions. Indeed, prior administration of rituximab and intensive chemotherapies, which are associated with increased PML risk,^15,23 were common in our cohort. Quantifying and comparing the attribution of each of these exposures is challenging. Nonetheless, the established immunosuppressive effect of CAR-T therapy,^4,5,24 coupled with increased PML reporting likelihood in the primary and most sensitivity analyses, support a meaningful contribution of CAR-T to PML risk.

In conclusion, post-marketing surveillance data suggest PML as a rare and potentially fatal complication of CAR-T therapy that may develop months to years later. Clinicians should be conscious of this complication, which is likely the result of prolonged B-cell aplasia and immunosuppression rather than direct neurological toxicity of CAR-T cells.