Risk of Natalizumab-Associated Progressive Multifocal Leukoencephalopathy

Transcription

1 T h e n e w e ngl a nd j o u r na l o f m e dic i n e original article Risk of Natalizumab-Associated Progressive Multifocal Leukoencephalopathy Gary Bloomgren, M.D., Sandra Richman, M.D., Christophe Hotermans, M.D., Meena Subramanyam, Ph.D., Susan Goelz, Ph.D., Amy Natarajan, M.S., Sophia Lee, Ph.D., Tatiana Plavina, Ph.D., James V. Scanlon, Pharm.D., Alfred Sandrock, M.D., and Carmen Bozic, M.D. A bs tr ac t From Biogen Idec, Weston, MA. Address reprint requests to Dr. Bloomgren at Biogen Idec, 14 Cambridge Center, Cambridge, MA 02142, or at gary.bloomgren@ biogenidec.com. N Engl J Med 2012;366: Copyright 2012 Massachusetts Medical Society. Background Progressive multifocal leukoencephalopathy (PML) is associated with natalizumab treatment. We quantified the risk of PML in patients with multiple sclerosis, according to the presence or absence of three risk factors: positive status with respect to anti JC virus antibodies, prior use of immunosuppressants, and increasing duration treatment. Methods We used data from postmarketing sources, clinical studies, and an independent Swedish registry to estimate the incidence of PML among natalizumab-treated patients with multiple sclerosis, according to positive or negative status with respect to anti JC virus antibodies, prior or no prior use of immunosuppressants, and duration of treatment (1 to 24 months vs. 25 to 48 months). Blood samples were available for anti JC virus antibody testing from 5896 patients with multiple sclerosis and from 54 patients with multiple sclerosis who were treated with natalizumab and in whom PML later developed. Results As of February 29, 2012, there were 212 confirmed cases of PML among 99,571 patients treated with natalizumab (2.1 cases per 1000 patients). All 54 patients with PML for whom samples were available before the diagnosis were positive for anti JC virus antibodies. When the risk of PML was stratified according to three risk factors, the risk of PML was lowest among the patients who were negative for anti JC virus antibodies, with the incidence estimated to be 0.09 cases or less per 1000 patients (95% confidence interval [CI], 0 to 0.48). Patients who were positive for anti JC virus antibodies, had taken immunosuppressants before the initiation of natalizumab therapy, and had received 25 to 48 months treatment had the highest estimated risk (incidence, 11.1 cases per 1000 patients [95% CI, 8.3 to 14.5]). Conclusions status with respect to anti JC virus antibodies, prior use of immunosuppressants, and increased duration treatment, alone or in combination, were associated with distinct levels of PML risk in natalizumab-treated patients with multiple sclerosis. (Funded by Biogen Idec and Elan Pharmaceuticals.) 1870

2 Natalizumab (Tysabri, Biogen Idec and Elan Pharmaceuticals) is approved for the treatment of relapsing remitting multiple sclerosis in more than 50 countries and also for the treatment of moderate-to-severe Crohn s disease in the United States. As of February 2012, approximately 100,000 patients have been treated with natalizumab worldwide (unpublished data). In the pivotal Natalizumab Safety and Efficacy in Relapsing Remitting Multiple Sclerosis trial (AFFIRM; ClinicalTrials.gov number, NCT ), natalizumab monotherapy decreased the risk of progression of disability by 42 to 54% and reduced the annualized rate of relapse by 68%. 1 Natalizumab treatment is associated with the risk of progressive multifocal leukoencephalopathy (PML), an opportunistic brain infection that is caused by the JC virus. PML appears to occur after a complex interaction between host and viral factors, leading to the development of a pathogenic form of JC virus that can infect and destroy oligodendrocytes in the central nervous system. 2,3 Natalizumab was voluntarily withdrawn from the market in 2005, after 3 cases of PML had been identified in clinical trials, representing an incidence of approximately 1 case per 1000 patients (95% confidence interval [CI], 0.2 to 2.8). 4-7 After the establishment of an intensive global riskmanagement program, natalizumab was reintroduced to the market in Accumulated safety data from this program have identified longer duration treatment, the use of immunosuppressants before the initiation therapy, and positive status with respect to anti JC virus antibodies, as assessed with the use of a two-step anti JC virus antibody assay (STRATIFY JCV, Focus Diagnostics), as risk factors for PML Descriptions of the risk among patients treated with natalizumab have so far been limited and mostly qualitative. 15,16 The purpose of this analysis was to provide updated estimates of the risk of PML among patients with multiple sclerosis according to these three risk factors. Me thods Patients and Data Collection Because of the infrequent occurrence of PML, data on natalizumab-treated patients with PML were collected from several sources, including postmarketing data from the Biogen Idec global safety database and from clinical studies as of February 29, Data used in this analysis were obtained from clinical trials and from spontaneous reporting. Patients enrolled in clinical studies provided informed consent for participation in those studies. For all other patients, data were collected through standard pharmacovigilance practices, for the purposes of monitoring the safety, as required by regulatory authorities, and informed consent was not required. Data on prior use of immunosuppressants were not available for all the patients exposed to natalizumab; therefore, we used the percentages of patients with and those without prior use of immunosuppressants in the Tysabri Global Observational Program in Safety study (TYGRIS, NCT [TYGRIS U.S., conducted in North America] and NCT [TYGRIS Rest of World]) as estimates for the overall natalizumab-treated population. TYGRIS is a multinational, observational cohort study designed to obtain long-term safety data on nataliz umab-treated patients with multiple sclerosis in a clinical practice setting. The estimate of the prevalence of anti JC virus antibodies in the general population of patients with multiple sclerosis was based on the prevalence among patients in three ongoing or completed clinical studies (AFFIRM, 1 STRATIFY-1 [NCT ], and TYGRIS-U.S.) and an independent Swedish registry of patients with multiple sclerosis, 17 as assessed on the basis of a single baseline plasma or serum sample obtained from each patient. In addition, the presence or absence of anti JC virus antibodies was assessed in patients treated with natalizumab in whom PML developed and from whom serum or plasma samples (or both) were obtained and stored before the diagnosis of PML, either as part of clinical studies or through spontaneous reports of PML in the postmarketing setting. All the authors vouch for the veracity and completeness of the data and analyses. Duration of Natalizumab Treatment as a Risk Factor for PML Estimates of the incidence of PML since the reintroduction into the market were made on the basis of the number of confirmed cases of PML at the time of this analysis and the postmarketing to natalizumab through February 29, We calculated the incidence of PML in each time period (12-month treatment in- 1871

3 T h e n e w e ngl a nd j o u r na l o f m e dic i n e tervals) by dividing the confirmed number of PML cases among patients treated with natalizumab during that time period by the total number of patients ever exposed to natalizumab during that same period. Prior Immunosuppressant Use as a Risk Factor for PML Data on the use of immunosuppressants before the initiation therapy among patients with multiple sclerosis in whom PML developed in the postmarketing period or during the course of clinical trials were obtained from the Biogen Idec global safety database and were compared with data on the prior use of immunosuppressants among patients with multiple sclerosis treated with natalizumab in the TYGRIS study. This analysis did not include data on the use of glucocorticoids, interferons, or glatiramer acetate. Anti JC Antibody Status as a Risk Factor for PML The overall prevalence of anti JC virus antibodies in the general population of patients with multiple sclerosis was estimated on the basis of the prevalence among patients in AFFIRM, 1 STRATIFY-1, TYGRIS-U.S., and an independent Swedish registry of patients with multiple sclerosis, 17 as assessed with the use of the two-step anti JC virus antibody assay, as described previously. 14 We also used this assay to determine the prevalence of anti JC virus antibodies before the development of PML among patients with multiple sclerosis treated with natalizumab (using all available samples that were obtained 6 months before the diagnosis of PML), and we compared this prevalence with that in the general population of patients with multiple sclerosis. Estimation of PML Incidence According to Anti JC Antibody Status The incidence of PML according to the presence or absence of anti JC virus antibodies was estimated as follows: the number of patients with multiple sclerosis in whom PML developed and who had samples available before the diagnosis of PML that tested positive for anti JC virus antibodies was divided by the estimated number -treated patients who were positive for anti JC virus antibodies; we used the same method to estimate the incidence of PML in patients who were negative for anti JC virus antibodies. The denominators were estimated on the basis of three factors: the number of patients with multiple sclerosis in whom PML developed and who had samples available before the diagnosis of PML that tested positive or negative for anti JC virus antibodies, the overall global postmarketing incidence of PML, and the prevalence of anti JC virus antibodies in the general population of patients with multiple sclerosis. To provide a conservative estimate of the incidence of PML in patients who were negative for anti JC virus antibodies, a sensitivity analysis was performed on the basis of the assumption of one hypothetical case of PML in a patient who was negative for anti JC virus antibodies; an additional sensitivity analysis was performed to assess this risk assuming 18 months or more therapy. Quantification of PML Risk According to Risk Factors A risk-factor algorithm was developed to estimate the incidence of PML among patients with and among patients without the established risk factors for natalizumab-associated PML. This algorithm estimated the risk of PML according to status with respect to anti JC virus antibodies (positive or negative), prior use of immunosuppressants (yes or no), and duration treatment (1 to 24 months vs. 25 to 48 months). First, we estimated the incidence of PML as follows: we determined the number of cases of PML among patients who had received 1 to 24 months treatment and among those who had received 25 to 48 months treatment, with or without the prior use of immunosuppressants; we then divided that number by the number -treated patients in the postmarketing setting who received at least 1 to 24 months treatment or at least 25 to 48 months treatment, stratified according to status with respect to the prior use of immunosuppressants (estimated by using the weighted proportion of prior immunosuppressant use in the TYGRIS-U.S. study and in the TYGRIS Rest of World study). Then, expanding on the two-risk-factor analysis, we estimated the incidence of PML among patients who were positive for anti JC virus antibodies by dividing the number of patients with natalizumab-associated PML who were positive for anti JC virus antibodies (the same numerator as in the two-factor analysis) by the estimated number of natalizu- 1872

4 mab-treated patients who were positive for anti JC virus antibodies in the postmarketing setting (the denominator in the two-factor analysis multiplied by the prevalence of anti JC virus antibodies observed in the general population of patients with multiple sclerosis in this analysis). In making this estimate, we assumed, for the numerator, that all confirmed cases -associated PML occurred in patients who were positive for anti JC virus antibodies before the diagnosis of PML, and we assumed, for the denominator, that the prevalence of anti JC virus antibodies in the general population of patients with multiple sclerosis in this analysis represented the percentage of natalizumab-treated patients who were positive for anti JC virus antibodies. The incidence of PML among patients who were negative for anti JC virus antibodies was derived from the sensitivity analysis that assumed one hypothetical case of PML in a patient who was negative for anti JC virus antibodies. statistical analysis The percentage of patients with detectable levels of anti JC virus antibodies in the general population of patients with multiple sclerosis and estimates of PML incidence were calculated along with two-sided exact 95% confidence intervals. A one-sided exact binomial test was used to compare the prevalence of anti JC virus antibodies in natalizumab-treated patients before the diagnosis of PML with the prevalence of anti JC virus antibodies in the general population of patients with multiple sclerosis in this analysis. A two-sided Fisher s exact test was used to compare the estimated incidence of PML in patients who were negative and in those who were positive for anti JC virus antibodies. R esult s Duration of Natalizumab Treatment as a Risk Factor for PML At the time of this analysis, 212 confirmed cases of PML had been identified globally; 46 of the affected patients (22%) died. In addition, 23 of the 58 survivors (40%) with available data on disability and 6 months or longer of follow-up after the diagnosis of PML had severe disability. A total of 99,571 patients had been exposed to natalizumab, representing 209,123 patient-years of experience (1.01 cases per 1000 patient-years). The risk of PML increased with increasing duration of treatment, with the greatest increase in risk occurring after 2 years of therapy (Fig. 1). Data beyond 4 years of therapy were limited. Prior Immunosuppressant Use as a Risk Factor for PML The use of immunosuppressants before the initiation therapy was more common among patients in whom PML subsequently developed than among patients enrolled in the TYGRIS study, which represented the overall patient population receiving natalizumab therapy. A total of 34.5% -treated patients with PML, as compared with 20.3% of all natalizumab-treated patients in the TYGRIS study (14.0% in the United States and 23.5% in the European Union), had received one or more immunosuppressants before the initiation therapy; the immunosuppressants used most frequently among natalizumab-treated patients with PML and among natalizumab-treated patients in the TYGRIS study included mitoxantrone, methotrexate, cyclophosphamide, azathioprine, and mycophenolate mofetil (Table 1). The duration of the prior use of immunosuppressants and the time from administration of the last dose of immunosuppressant to the start therapy varied widely among the patients with PML, and no patterns were observed. Anti JC Antibody Status as a Risk Factor for PML A total of 5896 patients from the AFFIRM, TYGRIS-U.S., and STRATIFY-1 studies and the Swedish multiple sclerosis registry had a baseline sample available for anti JC virus antibody testing. The demographic characteristics of the patients, including age and sex, and the percentage of patients who had taken immunosuppressants before the initiation therapy, when that information was available, were generally similar among these data sources when patients were stratified according to positive or negative status with respect to anti JC virus antibodies (Table 2). The prevalence of anti JC virus antibodies increased with age and was lower among women than among men, findings that are consistent with those in previous reports. 14,18 The prevalence of anti JC virus antibodies in this general population with multiple sclerosis was 54.9% (95% CI, 53.7 to 56.2). 1873

5 T h e n e w e ngl a nd j o u r na l o f m e dic i n e Incidence per 1000 Patients Clinical trials Postmarketing setting Duration of Natalizumab Treatment No. PML cases Patients treated , , , , , , Figure 1. Incidence of Progressive Multifocal Leukoencephalopathy (PML) Associated with Natalizumab Therapy, According to Duration of Treatment in 12-nth Intervals. Data on the incidence of PML in clinical trials are from Yousry and colleagues. 7 Estimates of the incidence of PML in the postmarketing setting were based on natalizumab and the 212 confirmed cases of PML as of February 29, The incidence of cases per 1000 patients in the postmarketing setting is derived from the 212 confirmed cases among 99,571 patients who received natalizumab therapy for at least 1 month. The risk of PML increased with increasing duration of treatment, with the greatest increase in risk occurring after 2 years of therapy (25 to 48 months). Data beyond 4 years of treatment were limited. I bars indicate 95% confidence intervals. One or more samples, obtained 6 to 187 months before the diagnosis of PML, were available from 54 natalizumab-treated patients with multiple sclerosis who were enrolled in premarketing or postmarketing clinical trials (31 patients) or who were identified from spontaneous reports of PML in the postmarketing setting (23 patients). The clinical characteristics of these 54 patients with PML were similar to those of the total cohort of patients with confirmed cases of PML (212 patients), indicating that there was no obvious selection bias (see Table 1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). All the samples from these 54 patients obtained at all time points including time points before the initiation treatment tested positive for anti JC virus antibodies (Fig. 1 in the Supplementary Appendix). In addition, 1 patient who had had a negative test for anti JC virus antibodies tested positive 13 months later, approximately 2 months before the onset of PML symptoms. The exact timing of the change from negative to positive status with respect to anti JC virus antibodies for this patient cannot be determined, but the change appears to have occurred in the interval between the two tests. Because this patient s positive test result occurred within 6 months before the diagnosis of PML and the serostatus between 6 and 15 months before the diagnosis of PML was unknown, this patient was excluded from the analysis. The 100% prevalence of positive status for anti JC virus antibodies before the diagnosis of PML among natalizumab-treated patients with multiple sclerosis in whom PML subsequently developed differed significantly from the estimated 54.9% prevalence of positive status for anti JC virus antibodies in the general population with multiple sclerosis (P<0.001), showing the usefulness of this two-step anti JC virus antibody assay for stratifying the risk of PML. Estimate of PML Incidence According to Anti JC Antibody Status The estimated incidence of PML differed significantly between patients who were positive and 1874

6 Table 1. Prior Use of Immunosuppressants among Patients with Multiple Sclerosis Treated with Natalizumab.* Variable Confirmed PML Cases in Postmarketing Setting among Patients with Prior Use of Immunosuppressants (N = 68) Patients in TYGRIS Study with Prior Use of Immunosuppressants (N = 792) Immunosuppressant used no. (%) Mitoxantrone 38 (56) 344 (43) Methotrexate 9 (13) 45 (6) Azathioprine 11 (16) 133 (17) Cyclophosphamide 14 (21) 71 (9) Mycophenolate mofetil 6 (9) 48 (6) Other 8 (12) 201 (25) Duration of prior immunosuppressant use mo Mean Median Range <1 24 Washout period between immunosuppressant and natalizumab mo Mean Median Range <1 24 * Included are data from natalizumab-treated patients with progressive multifocal leukoencephalopathy (PML) identified in the postmarketing setting and from natalizumab-treated patients enrolled in the Tysabri Global Observational Program in Safety (TYGRIS) study. Although data on any use of immunosuppressants (i.e., any use vs. no use during the patient s lifetime) before the initiation therapy were collected in the TYGRIS study, the dates treatment (which were used in the calculation of the duration of immunosuppressant use and the length of the washout period) were recorded only for immunosuppressants used in the 24 months preceding the initiation therapy. Patients may have received more than one type of prior immunosuppressant therapy. those who were negative for anti JC virus antibodies (P<0.001) (Table 3). We estimated the incidence of PML according to anti JC virus antibody status by means of the following calculation: on the basis of the overall incidence of PML of cases per 1000 patients worldwide who received natalizumab therapy for at least 1 month, it was estimated that the 54 patients with PML who had pre-pml samples available arose from approximately 25,364 patients receiving natalizumab treatment (i.e., [ ] = 25,364). It was then assumed that 55% of these 25,364 patients (13,950 patients) were positive for anti JC virus antibodies and 45% (11,414 patients) were negative for anti JC virus antibodies, as determined on the basis of the observed 55% prevalence of anti JC virus antibodies in the general population with multiple sclerosis in this analysis. Thus, the incidence of PML among patients positive for anti JC virus antibodies was estimated to be 3.87 cases per 1000 patients treated (95% CI, 2.91 to 5.05). Conversely, the estimated incidence of PML among patients negative for anti JC virus antibodies was 0 cases per 1000 patients (95% CI, 0 to 0.32). At the time of this analysis, the incidence of PML among patients who were negative for anti JC virus antibodies could not be fully ascertained, since no patient with PML had tested negative for anti JC virus antibodies without testing positive at a later date, before the diagnosis of PML. The true risk of PML among patients negative for anti JC virus antibodies cannot be zero, given that the anti JC virus antibody assay has an estimated analytical false negative rate between 2.5% and 2.7%. 13,14 Therefore, to estimate the incidence of PML among patients negative for anti JC virus antibodies, a sensitivity analysis was performed 1875

7 T h e n e w e ngl a nd j o u r na l o f m e dic i n e Table 2. Characteristics of the General Population with Multiple Sclerosis, According to Anti JC Antibody Status.* Variable AFFIRM (N = 823) TYGRIS-U.S. (N = 1480) STRATIFY-1 (N = 1096) Swedish Patients (N = 2497) Total (N = 5896) 2657 (45.1; ) 3239 (54.9; ) 1025 (41.0; ) 1472 (59.0; ) 482 (44.0; ) 614 (56.0; ) 776 (52.4; ) 704 (47.6; ) 374 (45.4; ) Patients no. (%; 95% CI) 449 (54.6; ) Age yr Mean Range Sex % Male Female Prior use of immunosuppressants % Yes NA NA No NA NA * The Natalizumab Safety and Efficacy in Relapsing Remitting Multiple Sclerosis trial (AFFIRM) was an international study. The TYGRIS-U.S. study is being conducted in the United States and Canada; information on age, sex, and prior use of immunosuppressants was available for 1451 of the 1480 patients. The STRATIFY-1 trial is being conducted in the United States; information on the prior use of immunosuppressants was available for 988 of the 1096 patients. The Swedish patients are a cohort from an independent Swedish registry of patients with multiple sclerosis; information on age was available for 2464 of the 2497 patients, and information on sex was available for 2494 patients. Information on the prior use of immunosuppressants was not available (NA) for the Swedish patients. in which 1 hypothetical case of PML in a patient who was negative for anti JC virus antibodies was assumed, allowing for determination of a conservative estimate that is, a point estimate that is probably higher than the true incidence. This analysis showed an estimated incidence of PML among patients negative for anti JC virus antibodies of 0.09 cases or less per 1000 patients (95% CI, 0 to 0.48), an incidence that is lower by a factor of at least 44 than the incidence among patients positive for anti JC virus antibodies (P<0.001); in addition, in a second sensitivity analysis performed on the basis of an assumption that all patients received at least 18 months treatment, the estimate of the incidence of PML among patients negative for anti JC virus antibodies was generally consistent with the estimate from the first sensitivity analysis (Table 3). Quantification of PML Risk According to the Three Risk Factors When the risk of PML was stratified according to anti JC virus antibody status (positive or negative), prior use of immunosuppressants (yes or no), and duration treatment (1 to 24 months or 25 to 48 months), five distinct subgroups of patients were identified (Fig. 2). Since to JC virus is a requirement for the development of PML, patients who were negative for anti JC virus antibodies represented the subgroup at the lowest risk in the PML risk-stratification algorithm, with an estimated incidence of 0.09 cases or fewer per 1000 patients (95% CI, 0 to 0.48), on the basis of the conservative estimate determined in the sensitivity analysis (Table 3). In the highest-risk subgroup (patients who had all three risk factors), the estimated incidence of PML was approximately 11.1 cases per 1000 patients (95% CI, 8.3 to 14.5). Discussion In this analysis, patients with multiple sclerosis were stratified into groups that were at a lower risk or a higher risk for PML on the basis of three risk factors. Since infection with JC virus is required for the development of PML, the risk of PML was lowest among patients who were negative for anti JC virus antibodies, with a conservatively estimated risk of 0.09 cases or fewer per 1000 natalizumab-treated patients (95% CI, 0 to 1876

8 Table 3. Estimated Incidence of PML According to Anti JC Antibody Status. Variable for Anti JC for Anti JC Total P Value Relative Risk of PML with Status Primary analysis* <0.001 (14.4 ) No. of PML cases Total patients treated 13,950 11,414 25,364 Incidence per 1000 patients 3.87 ( ) 0 ( ) 2.13 ( ) Sensitivity analysis I < ( ) No. of PML cases Total patients treated 14,209 11,625 25,834 Incidence per 1000 patients 3.80 ( ) 0.09 ( ) 2.13 ( ) Sensitivity analysis II < ( ) No. of PML cases Total patients treated ,764 Incidence per 1000 patients 6.23 ( ) 0.14 ( ) 3.49 ( ) * The primary analysis was based on the overall incidence of PML of cases per 1000 patients who received natalizumab therapy for at least 1 month. All 54 cases of PML occurred in patients who were positive for anti JC virus antibodies before the diagnosis of PML. Sensitivity analysis I was an analysis that was based on the assumption of one hypothetical patient with PML who was negative for anti JC virus antibodies; this assumption was made in order to derive a conservative estimate of the incidence of PML. Sensitivity analysis II was an analysis that was based on the assumption of one hypothetical patient with PML who was negative for anti JC virus antibodies. In addition, this analysis assumed that patients received natalizumab therapy for at least 18 months, since 53 of the 54 patients with PML received natalizumab for at least 18 months. 0.48). Conversely, the risk of PML was highest among patients who had all three risk factors (positive status with respect to anti JC virus antibodies, prior use of immunosuppressants, and natalizumab treatment for 25 to 48 months), with an estimated incidence of 11.1 cases per 1000 patients (95% CI, 8.3 to 14.5). The use of this algorithm may help identify patients for whom natalizumab therapy is most appropriate and may reduce the incidence of PML, a serious potential complication of therapy. The algorithm presented in this analysis, which can discern a difference by a factor of 120 in the risk of PML across the patient subgroups, allows for better characterization of risk than that observed with previous methods 7 and should assist clinicians in making more informed, individualized treatment decisions. Independent guidelines regarding testing for anti JC virus antibodies have begun to be disseminated. 19,20 Other factors to consider in determining whether natalizumab is an appropriate treatment include the severity of the multiple sclerosis in the patient, the patient s prior experience with disease-modifying drugs (e.g., neutralizing antibodies), the efficacy and safety of alternative therapies, and the manner in which risks are conveyed to the patient. Ultimately, the decisions regarding the initiation and duration therapy should be based on a discussion between the physician and the patient with consideration of all factors, including the findings of this analysis. This analysis has some limitations. First, data on long-term to natalizumab were limited, with approximately 46% of the patients having had 2 or more years of therapy and 14% having had 4 or more years of therapy (Fig. 1); for this reason, the analysis of the PML risk algorithm was confined to 48 months. Second, quantification of the risk of PML associated with prior use of immunosuppressants was based on data obtained from patients with confirmed PML 1877

9 T h e n e w e ngl a nd j o u r na l o f m e dic i n e A Prior immunosuppressant use? No Yes 1 24 nths nths 1 24 nths nths PML Cases (no.) Patients Treated (no.) PML Incidence per 1000 Patients 25 81, ( ) 94 37, ( ) 16 18, ( ) ( ) B Anti JC virus antibody status Prior immunosuppressant use? No Yes 1 24 nths nths 1 24 nths nths PML Cases (no.) Patients Treated (no.) PML Incidence per 1000 Patients 1 (hypothetical) 11, (0 0.48) 25 44, ( ) 94 20, ( ) 16 10, ( ) 52 4, ( ) Figure 2. Approximate Incidence of PML, Stratified According to Risk Factors. Estimates of the incidence of PML are shown, stratified according to prior or no prior use of immunosuppressants and duration treatment (Panel A) and according to positive or negative status with respect to anti JC virus antibodies, prior or no prior use of immunosuppressants, and duration treatment (Panel B). The number of cases of PML represents the number of confirmed cases of PML with the characteristic of interest (i.e., positive or negative status with respect to anti JC virus antibodies, prior or no prior use of immunosuppressants, and duration treatment of 1 to 24 months or 25 to 48 months) in the postmarketing setting among patients treated with natalizumab. At the time of the analysis, 175 confirmed cases of PML were documented in patients who had information available on the prior use of immunosuppressants and had received natalizumab treatment for 1 to 48 months. An additional 12 confirmed cases of PML were documented in patients who had received natalizumab for 1 to 48 months, but information on prior use of immunosuppressants was not available. We assumed that the percentage of patients with prior use of immunosuppressants among these patients was similar to that reported for the patients with PML who did have information, within each treatment interval (i.e., 34.5%). It was assumed that all patients with PML were positive for anti JC virus antibodies. In Panel A, the number of patients treated represents the number -treated patients in the postmarketing setting who received natalizumab treatment for at least 1 to 24 months or for at least 25 to 48 months, stratified according to prior or no prior use of immunosuppressants. In Panel B, the number of patients treated in the case of those positive for anti JC virus antibodies was estimated on the basis of the assumption that 55% of patients with multiple sclerosis (55% of the treated patients shown in Panel A) are positive for anti JC virus antibodies. The number of PML cases among those negative for anti JC virus antibodies represents one hypothetical case of PML in a patient who was negative for anti JC virus antibodies, and the number of patients treated represents the estimated number -treated patients who were negative for anti JC virus antibodies from the sensitivity analysis. 1878

10 and the subset -treated patients participating in the large, multinational TYGRIS study, with the assumption that the population enrolled in the TYGRIS study was representative of the overall natalizumab-treated population. Although information on the prior use of immunosuppressants is not obtained from all natalizumab-treated patients worldwide, the TYGRIS study includes the largest and most long-term data set available to date and probably represents the best estimate of the history of the use of immunosuppressants in natalizumab-treated patients. Recent data from the Tysabri Observational Program (TOP, NCT ), a smaller, long-term, safety follow-up study -treated patients, indicate that the percentage of patients with prior use of immunosuppressants is approximately 15%, a percentage that is slightly lower than, but relatively consistent with, data from the TYGRIS study (20.3%). 21 As with any clinical study in which medication history is recorded, recall bias may have affected the assessment of the prior use of immunosuppressants both among patients with PML and in the TYGRIS cohort. Third, the estimates of the risk of PML among patients who were positive for anti JC virus antibodies in the three-factor algorithm (Fig. 2B) were limited by the assumption that all confirmed cases of PML in the postmarketing setting occurred in patients who were positive for anti JC virus antibodies before the diagnosis of PML. This assumption was based on the fact that, at the time of this analysis, all 54 patients with PML for whom samples were available were positive for anti JC virus antibodies before PML was diagnosed; however, not all patients with PML had samples available for testing, and the anti JC virus antibody assay has a defined, albeit low, analytical false negative rate (2.5 to 2.7%). 13,14 The estimation of the incidence of PML among patients who were negative for anti JC virus antibodies was limited by the assumption of a hypothetical case of PML in a patient who was negative for anti JC virus antibodies. Finally, these estimates assumed a 55% prevalence of anti JC virus antibodies in the natalizumab-treated population with multiple sclerosis. Despite the potential for seroconversion in patients who are negative for anti JC virus antibodies and geographical differences in seroprevalence, current data indicate that the annual seroconversion rate is low (approximately 2%, which is similar to rates with other assays), and that the geographic variation in the seroprevalence of anti JC virus antibodies is minimal (Table 2). 14,18,22,23 Further data are needed to better characterize the magnitude of the risk of PML in patients who are positive and in those who are negative for anti JC virus antibodies. Despite these limitations, this analysis suggests that for patients with multiple sclerosis who are considering or receiving natalizumab therapy, the risk of PML can be stratified according to positive or negative status with respect to anti JC virus antibodies, prior or no prior use of immunosuppressants, and the duration treatment. Data from prospective studies are needed to further characterize these risks. All the authors report being employed by and holding stock in Biogen Idec. No other potential conflict relevant to this article was reported. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank numerous practicing neurologists for generously contributing pre-pml samples from natalizumab-treated patients with the goal of furthering PML research; and Tomas Olsson, M.D., Ph.D., for providing data on anti JC virus antibody testing from the independent Swedish registry. References 1. Polman CH, O Connor PW, Havrdova E, et al. A randomized, placebo-controlled trial for relapsing multiple sclerosis. N Engl J Med 2006;354: Major EO. Progressive multifocal leukoencephalopathy in patients on immunomodulatory therapies. Annu Rev Med 2010; 61: Tan CS, Koralnik IJ. Progressive multifocal leukoencephalopathy and other disorders caused by JC virus: clinical features and pathogenesis. Lancet Neurol 2010;9: Kleinschmidt-DeMasters BK, Tyler KL. Progressive multifocal leukoencephalopathy complicating treatment with natalizumab and interferon beta-1a for multiple sclerosis. N Engl J Med 2005;353: Langer-Gould A, Atlas SW, Green AJ, Bollen AW, Pelletier D. Progressive multifocal leukoencephalopathy in a patient treated with natalizumab. N Engl J Med 2005;353: Van Assche G, Van Ranst M, Sciot R, et al. Progressive multifocal leukoencephalopathy after natalizumab therapy for Crohn s disease. N Engl J Med 2005;353: Yousry TA, Major EO, Ryschkewitsch C, et al. Evaluation of patients treated with natalizumab for progressive multifocal leukoencephalopathy. N Engl J Med 2006; 354: Tysabri (natalizumab). Cambridge, MA: Biogen Idec, January 2012 (package insert). 9. Summary of product characteristics: Tysabri (natalizumab). Cambridge, MA: Biogen Idec, June Gorelik L, Bixler S, Lerner M, et al. Assessment of the incidence of anti-jcv antibodies in a cohort treated patients with multiple sclerosis. Mult Scler J 2010;16:Suppl:S306-S Olsson T, Hillert J, Alfredsson L, et al. Anti-JCV antibody prevalence in a Swedish cohort of MS patients and non-ms controls. Mult Scler J 2010;16:Suppl:S Subramanyam M, Plavina T, Simon K, et al. Factors associated with anti-jcv antibody prevalence in a large cohort of na- 1879

11 talizumab-treated MS patients. Mult Scler J 2010;16:Suppl:S38-S Bozic C, Richman S, Plavina T, et al. Anti-JCV antibody prevalence in multiple sclerosis patients: baseline results of STRATIFY-1. Ann Neurol 2011;70: Gorelik L, Lerner M, Bixler S, et al. Anti-JC virus antibodies: implications for PML risk stratification. Ann Neurol 2010; 68: Clifford DB, De Luca A, Simpson DM, Arendt G, Giovannoni G, Nath A. Natalizumab-associated progressive multifocal leukoencephalopathy in patients with multiple sclerosis: lessons from 28 cases. Lancet Neurol 2010;9: Vermersch P, Kappos L, Gold R, et al. Clinical outcomes -associated progressive multifocal leukoencephalopathy. Neurology 2011;76: Sverska MS-registret home page ( 18. Egli A, Infanti L, Dumoulin A, et al. Prevalence of polyomavirus BK and JC infection and replication in 400 healthy blood donors. J Infect Dis 2009;199: The Swedish MS Association home page ( 20. Warnke C, Adams O, Hartung HP, et al. Risk stratification of progressive multifocal leukoencephalopathy under natalizumab: recommendations for JC virus serology. Nervenarzt 2011;82: (In German.) 21. Butzkueven H, Polman C, Belachew S, et al. The incidence of PML in natalizumab long-term safety programs is comparable to postmarketing incidence. Mult Scler J 2011;17:Suppl:S Bozic C, Subramanyam M, Paes D, et al. JCV epidemiology in MS: epidemiology of anti-jc virus antibody prevalence in multiple sclerosis patients. Mult Scler J 2011; 17:Suppl:S355-S Olsson T, Achiron A, Alfredsson L, et al. Prevalence of anti-jcv antibodies in a multi-national cohort of multiple sclerosis patients. Mult Scler J 2011;17:Suppl:S134- S135. Copyright 2012 Massachusetts Medical Society. 1880