Abstract
Ruxolitinib is a JAK1/2 inhibitor approved for the treatment of myelofibrosis (MF) and polycythemia vera (PV). Recent data has suggested the possibility of increased infectious and secondary malignancy rates in patients treated with ruxolitinib. We conducted a dual-center, retrospective study of 202 myeloproliferative neoplasm (MPN) patients receiving ruxolitinib and a control cohort of 73 ruxolitinib-naïve MPN patients. We utilized propensity score matching to analyze the primary outcome of development of any grade infection. Infections occurred in 38.4% of ruxolitinib-naïve patients and 42.6% of ruxolitinib-treated patients and were primarily grade 1/2. After propensity score weighting, there was no difference in risk of infection between ruxolitinib treated and naïve patients with MF (HR 1.15 [95% CI 0.80 – 1.65], p=0.466) and non-MF MPNs (HR=0.52 [95% CI 0.21–1.28, p=0.152). These results suggest that there is not an increased risk of infection with ruxolitinib therapy.
Keywords: ruxolitinib, JAK inhibitor, infection, myeloproliferative
Introduction
Ruxolitinib (Jakafi, Incyte) is a Janus Associated Kinase (JAK) 1/2 inhibitor, FDA-approved for the treatment of intermediate and high-risk myelofibrosis (MF), as well as polycythemia vera (PV) with an inadequate response or intolerance to hydroxyurea (HU) [1]. Pivotal phase III studies have demonstrated the superiority of ruxolitinib over placebo and best available therapies in terms of disease-related symptom improvement and spleen volume reduction [2,3]. Ruxolitinib therapy is well tolerated overall, with predominantly predictable hematologic side effects including anemia and thrombocytopenia. However, as ruxolitinib was implemented into real-world practice, reports emerged noting an increased risk of infectious complications, including severe opportunistic infections, and secondary malignancies not fully appreciated in the prospective clinical trials. For instance, cases of cryptococcal meningitis and disseminated tuberculosis in patients with MF receiving ruxolitinib have been reported [4,5]. In addition, a report from a French Pharmacovigilance database detailed 30 cases of infections occurring in 26 patients treated with ruxolitinib, including several opportunistic infections [6].
Available evidence suggests that ruxolitinib may exert significant immunosuppressive activity by a variety of mechanisms. Ruxolitinib modulates dendritic cell function leading to impaired CD4+ and CD8+ T-cell priming and cytokine production. Most notably IL-12, a T-cell stimulating factor, is reduced via JAK1 inhibition. Ruxolitinib-mediated impairment of dendritic and T-cell response compounded with imbalances in cytokine production may drive the reactivation or acquisition of viral infections [7,8].
Porpazczy and colleagues recently assessed the incidence of secondary malignancies in 626 myeloproliferative neoplasms (MPN) patients treated with JAK1/2 inhibitors and found a statistically significant 16-fold increase in the risk of development of a B-cell malignancy, implicating JAK/STAT pathway inhibition in the emergence of C-MYC+ aggressive B-cell lymphomas [9]. In addition, a significant increase in non-melanoma skin cancer was noted in long term follow up of COMFORT-2 [10].
A recent meta-analysis assessed the additional infectious risk ruxolitinib may impart in the MPN patient population based on collated data from phase III randomized control trials and post-marketing surveillance, including case reports. The authors concluded there was insufficient evidence to estimate the risk of infection in ruxolitinib-treated patients [11]. Polverelli and colleagues examined a retrospective cohort of 507 MF patients at several centers in Italy to determine the epidemiology of infection and describe the impact of ruxolitinib treatment. Twenty-two percent of patients experienced 160 infection-related events, 45% of these events were graded as severe (grade 3–4). The rate of infection was statistically significantly higher in the ruxolitinib-treated cohort as compared to patients without ruxolitinib exposure (44% vs. 20%, p<0.001). International Prognostic Scoring System (IPSS) score of intermediate-2 or high risk and palpable splenomegaly greater than 10 cm below the left costal margin also emerged as risk factors for infection. Notably, both of these risk factors were more prevalent in the ruxolitinib-exposed cohort as compared to the ruxolitinib-naïve cohort [12]. This raises the possibility that disease-intrinsic factors, rather than ruxolitinib exposure, may mediate the observed increase in infectious complications. Comparison of the relative rates of infectious complications in MPN patients with and without ruxolitinib exposure is, therefore, necessary to further define the magnitude of risk.
Prospective evaluation utilizing a control cohort, while ideal, is currently not feasible given that ruxolitinib is the only FDA-approved drug for the treatment of MF and is widely prescribed in the community. We therefore sought to determine whether ruxolitinib increases risk of infection and occurrence of second malignancies using a large cohort of MPN patients with a propensity weighted control group of ruxolitinib naïve patients in a dual-center, retrospective study.
Methods
Patients
This was a retrospective, cohort study of patients with MPNs seen at two institutions: Mount Sinai Hospital (MSH) and Memorial Sloan Kettering Cancer Center (MSKCC) between 6/1/2010 and 12/31/2017. Patients were systematically identified through internal and external pharmacy records who had been treated with any dose of ruxolitinib for at least 7 days. Patients were excluded if they were less than 18 years of age, received ruxolitinib for another malignancy other than an MPN, or were primarily followed at an outside institution. A control cohort was identified at each institution consisting of MF and PV patients who were never treated with ruxolitinib between 1/1/1983 and 12/31/2017. The time period for the ruxolitinib naïve cohort was expanded given the high prevalence of ruxolitinib exposure in the MF population since FDA approval on November 6, 2011. This study was approved by the Institutional Review Board at both institutions.
Sample Size Calculation
A one-sided log-rank test accounting for competing risks was used to determine the target sample size of at least 185 patients (144 ruxolitinib-treated and 41 ruxolitinib-naive) needed to achieve 80% power at a 0.05 significance level to detect a hazard ratio of 1.836, deemed clinically meaningful based on prior retrospective data [12,13]. The 2-year cumulative incidence rate for infection was assumed to be 30% in the ruxolitinib-naive group and 50% in the ruxolitinib-treated group with a 2-year cumulative incidence rate for the competing risk of non-infection related deaths assumed to be 20% in the ruxolitinib-naive group and 10% in the ruxolitinib-treated group. The rate of loss to follow-up, defined as the percentage of patients who were alive but did not have infection within 2 years of starting ruxolitinib, was calculated using a sample dataset at Mount Sinai with patients who were eligible for the study with known infection rate estimated at 20%.
Definitions and Outcomes
Karyotype were deemed unfavorable if they included +8, −7/7q-, i(17q), −5/5q-, 12p-, inv(3), 11q23 or had three or more karyotypic abnormalities. Renal impairment was defined as creatinine clearance less than 60 mL/min and hepatic impairment was defined as a bilirubin greater than 2 mg/dL and an albumin less than 3.5 g/dL. Concomitant immunosuppressive medications included steroids (prednisone equivalent greater than 5mg daily) and chemotherapy, in addition to other medications with known immunosuppressive properties. Mutational information is reported, when available.
The primary outcome was time to development of any grade infection, graded by Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. Secondary outcomes were rates of hospitalization secondary to infection, time to diagnosis of secondary malignancy and time to death. Given difference in baseline infectious risk between MF versus PV patients, we separated the propensity score adjusted analysis into MF and non-MF cohorts.
Statistical Analysis
Distributions of continuous variables were summarized as mean (standard deviation) or median (interquartile range), as appropriate, and nominal variables as number (percentage).
Missing Data
Multiple imputations of missing data were conducted using the sequential regression method referred to as chained equations [14] implemented with the IVEware Version 0.3 software system [15]. Ten repetitions were performed to generate 10 imputed datasets.
Propensity Score Estimation
To account for selection bias, observed differences in baseline covariates, between ruxolitinib treated and ruxolitinib naïve groups, were subsequently controlled for with a weighted propensity score analysis using the inverse probability of treatment weighting (IPTW) method. Specifically, the probability of being in the two treatment groups was estimated from a multivariable logistic regression model, separately for the MF and non-MF populations, given their difference in risk factors. The variables included in the propensity score models were year of treatment (2011 or later), age at treatment, age at diagnosis, Dynamic International Prognostic Scoring System (DIPSS) category, and prior lines of therapy for the MF group and age at diagnosis, age at treatment, and prior lines of therapy for the non-MF group. The weights were calculated to estimate the average treatment effect among the treated (ATT) with ruxolitinib and ruxolitinib naïve control groups weighted so that the distribution of baseline covariates in each of the two samples was the same as in the sample of patients receiving ruxolitinib. Standardized mean differences in the weighted populations were computed to assess balance of baseline covariates with differences larger than 0.20 indicative of residual imbalance.
Survival Modeling
A competing risks analysis was used to model times to infection and second malignancy, accounting for death as a competing event, as described by Fine and Gray, weighted by the ATT- inverse probability of treatment [16]. A Cox proportional hazards regression analysis was used to model overall survival (OS), weighted by the ATT-inverse probability of treatment.
To account for multiple imputation of missing data, ATT-IPTW competing risk analyses and ATT-IPTW cox proportional hazards regression analyses were performed on each of the 10 imputed datasets, and estimates of corresponding hazard ratios (HRs) and confidence intervals (CIs) were then appropriately combined using the MIANALYZE procedure in SAS [17].
All analyses were performed using SAS Enterprise Guide 7.15 (SAS institute Inc, Cary, NC). Hypothesis testing was two-sided and conducted at the 5% level of significance.
Results
Patient Characteristics
A total of 202 MPN patients treated with ruxolitinib were identified, 153 (75.8%) with MF, 31 (15.3%) with PV and 18 (9.0%) with unspecified or other MPN. A control cohort of 73 ruxolitinib naïve MPN patients was identified, including 57 patients (78.1%) with MF, and 14 patients (19.2%) with PV. The types of therapies administered to these patients are shown in Table 2, and include no pharmacologic intervention (n=26), hydroxyurea (n=18), and peginterferon alfa-2a (n=5). The median duration of ruxolitinib treatment in the MF cohort was 19.03 months (interquartile range 9.03–34.23) and in the non-MF cohort it was 15.1 months (interquartile range 8.30–29.4).
Table 2.
Ruxolitinib Naïve Control Cohort Therapies
| Myelofibrosis | N=57 |
|---|---|
| No pharmacological intervention | 26 (45.6%) |
| Hydroxyurea | 8 (14%) |
| Anagrelide | 3 (5.3%) |
| Danazol | 3 (5.3%) |
| Peginterferon alfa-2a | 3 (5.3%) |
| Darbepoetin alfa | 2 (3.5%) |
| Decitabine | 2 (3.5%) |
| LBH589 | 2 (3.5%) |
| Thalidomide | 2 (3.5%) |
| Busulfan | 1 (1.8%) |
| KB004 | 1 (1.8%) |
| Luspatercept | 1 (1.8%) |
| Melphalan | 1 (1.8%) |
| Pomalidomide | 1 (1.8%) |
| PRM-151 | 1 (1.8%) |
| Non-myelofibrosis | N=16 |
| Hydroxyurea | 10 (62.5%) |
| Peginterferon alpha-2a | 4 (25%) |
| Anagrelide | 2 (12.5%) |
Baseline characteristics for the ruxolitinib-treated and -naïve cohorts are described in Table 1. As noted, the ruxolitinib cohort was significantly older at treatment initiation (68 vs. 64 years; p=0.003) and had received more lines of prior therapy with 70% of ruxolitinib treated patients vs. 30% of ruxolitinib naïve patients receiving at least 1 prior line of therapy (p<0.001). The ruxolitinib cohort had a higher rate of neutropenia (defined as absolute neutrophil count below 1000 cells/microL) at treatment initiation as compared to the ruxolitinib naïve control cohort (5% vs. 0%, p=0.015). In addition, the ruxolitinib cohort had a higher prevalence of concomitant immunosuppressive medications (24.8% vs. 4.1%, p<0.001).
Table 1.
Baseline Demographics
| Ruxolitinib naïve (N=73) | Ruxolitinib treated (N=202) | Total (N=275) | P value | |
|---|---|---|---|---|
| Age at treatment, years (median (IQR)) | 64 (51–71) | 68 (59–76) | 67 (57–75) | 0.003 |
| Female, N(%) | 34 (46.6%) | 96 (47.5%) | 130 (47.3%) | 0.889 |
| Diagnosis, N(%) | ||||
| Primary Myelofibrosis | 28 (38.4%) | 72 (35.6%) | 100 (36.4%) | 0.375 |
| Secondary Myelofibrosis | 29 (39.7%) | 81 (40.1%) | 110 (40.0%) | |
| Polycythemia Vera | 14 (19.2%) | 31 (15.3%) | 45 (16.4%) | |
| Unspecified MPN | 0 (0.0%) | 9 (4.5%) | 9 (3.3%) | |
| Other | 2 (2.7%) | 9 (4.5%) | 11 (4.0%) | |
| Unfavorable Karyotype, N(%) | 6 (8.2%) | 15 (7.4%) | 21 (7.6%) | 0.827 |
| Mutations, N (%) | ||||
| JAK2 | 45 (61.6%) | 161 (79.7%) | 206 (74.9%) | 0.002 |
| CALR | 18 (24.7%) | 14 (6.9%) | 32 (11.6%) | <.001 |
| MPL | 3 (4.1%) | 7 (3.5%) | 10 (3.6%) | 0.801 |
| None | 5 (6.8%) | 15 (7.4%) | 20 (7.3%) | 0.871 |
| Non-driver mutations | 17 (23.3%) | 17 (8.4%) | 34 (12.4%) | <.001 |
| DIPSS Category, N(%)* | ||||
| Low-risk | 15 (26.3%) | 13 (8.5%) | 28 (13.3%) | <.001 |
| Intermediate-1 risk | 21 (36.8%) | 43 (28.1%) | 64 (30.5%) | |
| Intermediate-2 risk | 17 (29.8%) | 60 (39.2%) | 77 (36.7%) | |
| High risk | 4 (7.0%) | 24 (15.7%) | 28 (13.3%) | |
| Unknown | 0 (0.0%) | 13 (8.5%) | 13 (6.2%) | |
| Baseline renal impairment, N(%) | 10 (13.7%) | 51 (25.2%) | 61 (22.2%) | 0.042 |
| Baseline hepatic impairment, N(%) | 1 (1.4%) | 5 (2.5%) | 6 (2.2%) | 0.58 |
| Prior non-MPN Malignancy, N(%) | 13 (17.8%) | 31 (15.3%) | 44 (16.0%) | 0.623 |
| Concomitant Immunosuppressive medications | 3 (4.1%) | 50 (24.8%) | 53 (19.3%) | <.001 |
| Prior Treatment Lines of Therapy | <.001 | |||
| 0 | 51 (69.9%) | 60 (29.7%) | 111 (40.4%) | |
| 1 | 15 (20.5%) | 85 (42.1%) | 100 (36.4%) | |
| 2 | 3 (4.1%) | 35 (17.3%) | 38 (13.8%) | |
| 3 | 2 (2.7%) | 13 (6.4%) | 15 (5.5%) | |
| 4 | 1 (1.4%) | 8 (4.0%) | 9 (3.3%) | |
| Unknown | 0 (0.0%) | 2 (1.0%) | 0 (0.0%) |
DIPSS only calculated for MF patients
Incidence of Infectious Complications and Secondary Malignancies
The median duration of follow-up of the ruxolitinib cohort was 21 months (range 1 to 84 months) and 44 months (range 1 to 353 months) for the ruxolitinib naive cohort. Table 3 shows the types and grades of infections that occurred in each cohort. The incidence of any grade infection in the ruxolitinib cohort was 42.6% and it was 38.4% in the ruxolitinib naïve cohort (p=0.531). Upper respiratory infections were most common in both cohorts, followed by urinary tract infection and pneumonia. Herpes zoster infection occurred in 3.9% of patients in the ruxolitinib cohort and 2.7% of patients in the ruxolitinib naïve control cohort (p=0.631). There was no difference in infection grades between the ruxolitinib treated and ruxolitinib naïve cohorts in this unadjusted analysis. There was also no difference between rates of grade 3 or higher infections. Of note, there was no difference between institutions in the combined cohort for types of infection (p=0.794) or frequency (p=0.986).
Table 3.
Infection Outcomes
| Infection Type | Ruxolitinib naïve (N=73) | Ruxolitinib treated (N=202) | Total (N=275) | P Value |
|---|---|---|---|---|
| Infection, Yes, N(%) | 28 (38.4%) | 86 (42.6%) | 114 (41.5%) | 0.531 |
| Respiratory | ||||
| Viral upper respiratory infection | 12 (16.4%) | 26 (12.8%) | 38 (13.8%) | |
| Pneumonia, bacterial | 0 (0%) | 12 (5.9%) | 12 (4.4%) | |
| Pneumonia, atypical | 0 (0%) | 2 (0.9%) | 2 (0.7%) | |
| Influenza | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Pulmonary tuberculosis | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Urinary | ||||
| Cystitis | 5 (6.8%) | 9 (4.4%) | 14 (5.1%) | |
| Pyelonephritis | 1 (1.3%) | 0 (0%) | 1 (0.4%) | |
| Skin and Soft Tissue | ||||
| Herpes zoster | 2 (2.7%) | 8 (3.9%) | 10 (3.6%) | |
| Cellulitis | 2 (2.7%) | 2 (0.9%) | 4 (1.5%) | |
| Herpes simplex virus | 0 (0%) | 2 (0.9%) | 2 (0.7%) | |
| Skin abscess | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Fungal rash | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Gastrointestinal | ||||
| Clostridium difficile | 0 (0%) | 4 (1.9%) | 4 (1.5%) | |
| Diverticulitis | 1 (1.3%) | 3 (1.4%) | 4 (1.5%) | |
| Enteritis | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Proctitis | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Norovirus | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Oral, Ocular, and Sinuses | ||||
| Dental abscess | 2 (2.7%) | 4 (1.9%) | 6 (2.2%) | |
| Bacterial conjunctivitis | 0 (0%) | 2 (0.9%) | 2 (0.7%) | |
| Sinusitis | 1 (1.3%) | 1 (0.4%) | 2 (0.7%) | |
| Otitis media | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Mucositis | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Miscellaneous | ||||
| Bacteremia | 1 (1.3%) | 3 (1.4%) | 4 (1.5%) | |
| Septic shock (unknown source) | 1 (1.3%) | 1 (0.4%) | 2 (0.7%) | |
| Osteomyelitis | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Lyme disease | 0 (0%) | 1 (0.4%) | 1 (0.4%) | |
| Infection Grade, N(%) | 0.813 | |||
| 1 | 5 (6.8%) | 9 (4.5%) | 14 (5.1%) | |
| 2 | 16 (21.9%) | 46 (22.8%) | 62 (22.5%) | |
| 3 | 6 (8.2%) | 26 (12.9%) | 32 (11.6%) | |
| 4 | 1 (1.4%) | 4 (2.0%) | 5 (1.8%) | |
| 5 | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Hospitalization Secondary to Infection, N(%) | 7 (9.6%) | 30 (14.9%) | 37 (13.5%) | 0.259 |
| Documented Recurrent Infection, N(%) | 1 (1.4%) | 29 (14.4%) | 30 (10.9%) | 0.002 |
There was no significant difference in secondary malignancies between the ruxolitinib treated and naïve cohorts in an unadjusted analysis (Table 4). The types of secondary malignancies are shown in Table 3. Skin cancers were the most common; no patients had documented precancerous lesions. In terms of prior hydroxyurea use, one skin cancer patient in the ruxolitinib naïve cohort had 15 years of hydroxyurea treatment and two in the ruxolitinib cohort had a history of hydroxyurea therapy, with 4 and 20 years of exposure. Notably, one patient developed diffuse large B-cell lymphoma (DLBCL) and one patient developed multiple myeloma in the ruxolitinib cohort.
Table 4.
Secondary Malignancy Outcomes
| Variable | Ruxolitinib naïve (N=73) | Ruxolitinib Treated (N=202) |
Total (N=275) | P Value |
|---|---|---|---|---|
| Secondary Malignancy, Yes, N(%) | 4 (5.5%) | 9 (4.5%) | 13 (4.7%) | 0.724 |
| Type of Cancers, N(%) | 0.733 | |||
| Squamous cell carcinoma | 1 (1.4%) | 3 (1.5%) | 3 (1.1%) | |
| Basal cell carcinoma | 1 (1.4%) | 1 (0.5%) | 1 (0.4%) | |
| Bladder cancer | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Diffuse large B-cell lymphoma | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Hodgkin’s lymphoma | 1 (1.4%) | 0 (0.0%) | 1 (0.4%) | |
| Lung adenocarcinoma | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Multiple myeloma | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Rectal carcinoma | 0 (0.0%) | 1 (0.5%) | 1 (0.4%) | |
| Pleomorphic carcinoma of lung | 1 (1.4%) | 0 (0.0%) | 1 (0.4%) | |
| Acute Leukemia, Yes, N(%) | 2 (2.7%) | 18 (8.9%) | 20 (7.3%) | 0.082 |
Propensity Score Weighted Survival Analysis
Given significant differences in baseline covariates between the ruxolitinib treated and naïve cohorts, a propensity score analysis was performed. The baseline characteristics before and after propensity score weighting for MF and non-MF patients are shown in Table S1 and Table S2, respectively. The standardized difference was reduced across nearly all variables in all categories, although some variables were still not well balanced with absolute standardized differences greater than 0.2. To account for residual imbalance, all variables were additionally included in multivariable competing risk and Cox models as covariates.
Table 5 shows the HR for risk of infection and second malignancy in the MF and non-MF cohort. For risk of infection, a propensity score weighted competing risk analysis did not reveal a statistically significant difference between the ruxolitinib treated and naïve cohorts in either the MF (HR 1.15 [95% CI 0.80 – 1.65], p=0.466) or non-MF populations (HR 0.52 [95% CI 0.21–1.28], p=0.152).
Table 5.
Propensity Score Weighted Analysis for Risk of Infections and Second Malignancy with Ruxolitinib Treatment
| Hazard ratio (95% CI) | P value | |
|---|---|---|
| Risk of Infection | ||
| Myelofibrosis Cohort | 1.15 (0.80 – 1.65) | 0.466 |
| Non-myelofibrosis Cohort | 0.52 (0.21 – 1.28) | 0.152 |
| Risk of Second Malignancy | ||
| Myelofibrosis Cohort* | NE | NE |
| Non-myelofibrosis Cohort | 10.81 (2.54 – 45.92) | 0.001 |
Given low event rate, a model was not able to be constructed
NE = not evaluable
For risk of second malignancy, a low event rate in the MF population prevented a model from being constructed. In the non-MF population, a propensity score weighted competing risk analysis revealed a statistically significant increased risk of second malignancy (HR 10.81 [95% CI 2.54 −45.92], p=0.001) with ruxolitinib treatment, however, with only two secondary malignancies in each group this result should be interpreted with caution.
There was no significant difference in OS between the ruxolitinib treated and naïve groups in the MF population (HR 1.43 [95% CI 0.68–2.99], p=0.345). The non-MF population did not have a sufficient number of events in the ruxolitinib naïve cohort to perform a survival analysis.
Discussion
Based on anecdotal clinical experience, we initially hypothesized that ruxolitinib treatment was associated with an increased risk of infection. However, the results of this study failed to show any significant difference in rates of infections between ruxolitinib treated patients and a ruxolitinib naïve cohort. Additionally, a low secondary malignancy event rate precluded the ability to distinguish a meaningful difference between these two cohorts of patients.
Much of the available literature regarding ruxolitinib-related infections relies on post-marketing studies without a control cohort [12,18–22]. Therefore, it is difficult to estimate the additional infectious risk associated with ruxolitinib treatment in the MPN population. In this propensity score analysis, ruxolitinib treatment was not significantly associated with an increased risk of infection as compared to a control cohort of ruxolitinib naïve patients in both the MF and non-MF sub-groups.
Infections were not systematically reported in the original publications of phase III studies of ruxolitinib [2,3,23]. In extended phase publications, only herpes zoster showed a statistically significant odds ratio with ruxolitinib treatment [11]. Notably, the incidence of serious pneumonia, fatal staphylococcal infections, fatal sepsis, grade 3/4 infections, or any grade infection were not significantly different between the ruxolitinib and control arm [11]. In the current study reported here, there was a numerically higher rate of herpes zoster infection associated with ruxolitinib treatment (3.9% vs. 2.7%), although this was not statistically significant (p=0.631). This may be due to a low event rate and relatively small sample size.
Our method of propensity score weighting in this study is important as patients with MF are at baseline risk of developing infections. Prior to the introduction of ruxolitinib, a large Swedish study examined a cohort of MPN patients, concluding MPN patients had a higher mortality rate than that of age and sex matched controls, primarily due to death as a direct result of the hematologic malignancy and infections. Hazard ratio (HR) of death from hematologic malignancies and infections were 92.8 (95% CI, 70.0 to 123.1) and 2.7 (95% CI, 2.4 to 3.1), respectively. Patients with MF were at particularly increased risk, with a 10.4% ten-year probability of dying from an infection in male patients aged 70–79 years [24]. Therefore, patients with MPNs, in particular MF, are at a baseline increased risk of death from infection.
Our cohort had a low event rate of secondary non-hematologic malignancies during the follow-up period (4.7% of the total cohort). Therefore, it was not possible to perform regression analysis, however, there was no difference in unadjusted rates (Table 3). The spectrum of types of second malignancies that occurred included several patients with squamous cell carcinoma of the skin as well as one patient with multiple myeloma and a patient with DLBCL. The latter finding is particularly relevant given a recent report of increased rate of B-cell malignancies in patients treated with ruxolitinib [9]. However, a database review did not find a significant difference in the incidence of a subsequent lymphoma diagnosis in MPN patients when comparing those who received prior JAK inhibitor therapy versus those who did not [25].
This study has a number of limitations that are important to recognize. There was difficulty accurately capturing infectious outcomes in a retrospective fashion, as diagnosis and grading data was only available per provider documentation. Additionally, patients may have had infectious complications that were either not reliably reported or faithfully reflected in the medical chart. The duration of ruxolitinib exposure may have also been inadequate to detect differences in the risk of infections or malignancy. Most notably, our ruxolitinib naïve control cohort was limited in size, reducing the ability to construct a well-matched cohort. Given the high prevalence of ruxolitinib exposure, it was challenging to identify ruxolitinib naïve subjects with sufficient extractable data to include in this study. Despite these limitations, this is among the largest retrospective cohort studies to examine the association between ruxolitinib exposure and risk of infection.
The diversity of treatments in the ruxolitinib naïve cohort, including nearly 40% not receiving active pharmacologic treatment, also limited matched comparisons with the ruxolitinib treated cohort. However, the ruxolitinib naïve cohort was significantly younger, had lower DIPSS scores, less neutropenia, and less prior lines of therapy. These patient characteristics would presumably influence the infectious risk with an expectation of lower incidence in the ruxolitinib naïve cohort than the ruxolitinib treated cohort in which more advanced disease features were present. The fact that a difference in infectious risk was not detected is perhaps even more convincing that no true difference in risk exists.
In conclusion, in this propensity score weighted retrospective study, ruxolitinib treatment was not associated with an increased risk of infection compared to ruxolitinib naïve patients. In an unadjusted analysis, a difference in occurrence of secondary malignancies was not documented.
This is the most methodologically rigorous retrospective study to examine the concern of ruxolitinib associated infectious risk and is the only study we are aware of that utilizes a propensity score weighted control cohort, thereby, minimizing a major limitation of other retrospective studies. Given that a prospective, control-cohort study is unlikely to be performed, larger propensity score weighted retrospective analyses should be done to confirm these findings. Identifying the actual risk of developing infections in the setting of ruxolitinib will only have clinical significance if a risk based antimicrobial prophylaxis strategy can be effectively employed and alternative active agents utilized to avoid secondary malignancy in those in which the risk outweighs the benefit of JAK inhibition.
Supplementary Material
Disclosure of Interest
M.K. has received research funding from Incyte, Celgene, Constellation, and Blueprint Medicines, and consulting fees from La Jolla Pharmaceutical. M.M has served as a consultant for Bristol-Myers Squibb, Novartis, Pfizer, and Takeda and has received research support to his institution from Bristol-Myers Squibb and SPARC/Sun Pharma. R.H. serves on the advisory board Novartis and La Jolla Pharmaceuticals. R.K.R has received consulting fees from Incyte, Celgene, Agios Pharmaceuticals, Apexx Oncology, and Jazz Pharmaceuticals and has received research funding from Constellation Pharmaceuticals, Incyte, and Stemline Therapeutics. J.M. has received clinical research funding to the institution from Incyte, CTI Biopharma, Janssen, Merck, Roche, Promedior, and Celgene. Clinical trials steering committee membership for Incyte, CTI Biopharma, Roche, Celgene, and Pharmaessentia.
References
- 1.United States Food and Drug Administration. JAKAFI (Ruxolitinib) Label. 2011.
- 2.Verstovsek S, Mesa RA, Gotlib J, et al. . A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med 2012;366:799–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Harrison C, Kiladjian JJ, Al-Ali HK, et al. . JAK inhibition with ruxolitinib versus best available therapy for myelofibrosis. N Engl J Med 2012;366:787–798. [DOI] [PubMed] [Google Scholar]
- 4.Chen CC, Chen YY, Huang CE. Cryptococcal meningoencephalitis associated with the long-term use of ruxolitinib. Ann Hematol 2016;95:361–362. [DOI] [PubMed] [Google Scholar]
- 5.Branco B, Metsu D, Dutertre M, et al. . Use of rifampin for treatment of disseminated tuberculosis in a patient with primary myelofibrosis on ruxolitinib. Ann Hematol 2016;95:1207–1209. [DOI] [PubMed] [Google Scholar]
- 6.Sylvine P, Thomas S, Pirayeh E, French Network of Regional Pharmacovigilance C. Infections associated with ruxolitinib: study in the French Pharmacovigilance database. Ann Hematol 2018;97:913–914. [DOI] [PubMed] [Google Scholar]
- 7.Heine A, Held SA, Daecke SN, et al. . The JAK-inhibitor ruxolitinib impairs dendritic cell function in vitro and in vivo. Blood 2013;122:1192–1202. [DOI] [PubMed] [Google Scholar]
- 8.Heine A, Brossart P, Wolf D. Ruxolitinib is a potent immunosuppressive compound: is it time for anti-infective prophylaxis? Blood 2013;122:3843–3844. [DOI] [PubMed] [Google Scholar]
- 9.Porpaczy E, Tripolt S, Hoelbl-Kovacic A, et al. . Aggressive B-cell lymphomas in patients with myelofibrosis receiving JAK1/2 inhibitor therapy. Blood 2018;132:694–706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Harrison CN, Vannucchi AM, Kiladjian JJ, et al. . Long-term findings from COMFORT-II, a phase 3 study of ruxolitinib vs best available therapy for myelofibrosis. Leukemia 2016;30:1701–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lussana F, Cattaneo M, Rambaldi A, Squizzato A. Ruxolitinib-associated infections: A systematic review and meta-analysis. Am J Hematol 2018;93:339–347. [DOI] [PubMed] [Google Scholar]
- 12.Polverelli N, Breccia M, Benevolo G, et al. . Risk factors for infections in myelofibrosis: role of disease status and treatment. A multicenter study of 507 patients. Am J Hematol 2017;92:37–41. [DOI] [PubMed] [Google Scholar]
- 13.Ritchie EK, Krichevsky S, Roboz GJ, et al. . Incidence of Infections and Second Cancers in Philadelphia Chromosome-Negative Patients with Myeloproliferative Neoplasms Treated with Ruxolitinib. American Society of Hematology Annual Meeting. Volume 130; 2017. [Google Scholar]
- 14.White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med 2011;30:377–399. [DOI] [PubMed] [Google Scholar]
- 15.Raghunathan T, Solenberger P, Hoewyk J. IVEware Imputation and Variance Estimation Software 2007.
- 16.Fine J, Gray R. A proportional hazards model for the subdistribution of a competing risk. Journal of the American Statistical Association 1999;94:496. [Google Scholar]
- 17.Yuan YC. Multiple Imputation for Missing Data: Concepts and New Developments. 2000; Cary, NC. [Google Scholar]
- 18.Al-Ali HK, Griesshammer M, le Coutre P, et al. . Safety and efficacy of ruxolitinib in an open-label, multicenter, single-arm phase 3b expanded-access study in patients with myelofibrosis: a snapshot of 1144 patients in the JUMP trial. Haematologica 2016;101:1065–1073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Breccia M, Andriani A, Montanaro M, et al. . Ruxolitinib in clinical practice for primary and secondary myelofibrosis: an analysis of safety and efficacy of Gruppo Laziale of Ph-negative MPN. Ann Hematol 2017;96:387–391. [DOI] [PubMed] [Google Scholar]
- 20.Chen YY, Huang CE, Lee KD, Chen CC. Clinical efficacy and safety of ruxolitinib in the management of myelofibrosis: A single institution experience in Taiwan . Hematology 2016;21:3–9. [DOI] [PubMed] [Google Scholar]
- 21.Ellis MH, Lavi N, Mishchenko E, et al. . Ruxolitinib treatment for myelofibrosis: Efficacy and tolerability in routine practice. Leuk Res 2015. [DOI] [PubMed] [Google Scholar]
- 22.Komatsu N, Kirito K, Shimoda K, et al. . Assessing the safety and efficacy of ruxolitinib in a multicenter, open-label study in Japanese patients with myelofibrosis. Int J Hematol 2017;105:309–317. [DOI] [PubMed] [Google Scholar]
- 23.Vannucchi AM, Kiladjian JJ, Griesshammer M, et al. . Ruxolitinib versus standard therapy for the treatment of polycythemia vera. N Engl J Med 2015;372:426–435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hultcrantz M, Wilkes SR, Kristinsson SY, et al. . Risk and Cause of Death in Patients Diagnosed With Myeloproliferative Neoplasms in Sweden Between 1973 and 2005: A Population-Based Study. J Clin Oncol 2015;33:2288–2295. [DOI] [PubMed] [Google Scholar]
- 25.Pemmaraju N, Kantarjian H, Nastoupil L, et al. . Characteristics of patients with myeloproliferative neoplasms with lymphoma, with or without JAK inhibitor therapy. Blood 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.