Withdrawal of ruxolitinib increases risk of mortality in patients with myelofibrosis and COVID-19

Outcomes of coronavirus 2019 disease (COVID-19) in patients with MPN are still not fully understood but the need for evidence-based recommendations on management of antithrombotic and cytoreductive therapies in patients infected with COVID-19 is urgent. In August 2020, a preliminary analysis of the European LeukemiaNet (ELN) observational study of COVD-19 in patients with MPN was presented by the MPN Hub Steering Committee member Tiziano Barbui and his colleagues, at the Texas Virtual MPN Workshop (reviewed on the MPN Hub here), which showed an increased incidence of thromboembolic events and high mortality rates especially in patients with myelofibrosis (MF).¹

An approved treatment for MF and polycythemia vera (PV) is the Janus kinase (JAK)1 and JAK2 inhibitor ruxolitinib. The anti-inflammatory and immunomodulatory properties of ruxolitinib have led investigators to assess it as a potential treatment for severe COVID-19; however, results have been disappointing (see article on the MPN Hub here).

The first results of an observational study of the ELN, assessing the impact of ongoing ruxolitinib treatment in patients with MPN who develop COVID-19, was recently published by Tiziano Barbui and co-authors in Leukemia, which we have summarized below.

Study design

This multicenter, observational, retrospective study analysed 180 adults with MPN treated in 38 centers across Europe (Italy, Spain, UK, France, Poland, and Germany).

Data were collected between February 15 and May 31, 2020, and included demographics, comorbidities, medications, laboratory tests, imaging, and treatments and supportive interventions for COVID-19. The median follow-up was 50 days. Additionally, information was collected on MPN driver mutations, duration of MPN prior to COVID-19 infection, presence of splenomegaly, and use of cytoreductive drugs.

Patient characteristics

• The median age was 71 years and 58.3% were male
• The majority (88%) of patients came from Italy and Spain
• Most patients were diagnosed with MF (34.3%), followed by essential thrombocytosis (ET; 29.1%), PV (26.3%) and prefibrotic primary MF (pre-PMF; 10.3%)
• The most common driver mutation was JAK2^V617F (72.4%) while mutations in CALR (28.4%) and MPL (5.6%) were less frequent
• 26.8% of patients had palpable splenomegaly. Common comorbidities seen in > 20% of the patients were hypertension (60.8%), smoking (23.0%), and hyperlipidemia (28.0%)

COVID-19 management

• Most COVID-19 infections in the cohort were diagnosed by a positive real-time reverse transcriptase polymerase chain reaction (RT-PCR) from nasal swabs
• In half of the patients managed at home (11% of the total population) diagnosis was based only on presence of pneumonia and symptoms highly suggestive of COVID-19 (e.g., fever, cough, oxygen saturation ≤ 93% at rest, dyspnea, diarrhea) due to the limited availability of tests
• Hospitalization was required for 77% of patients with 66.3% of them being managed on regular wards, and 10.9% in intensive care units (ICU)
• 47.7% of patients required noninvasive respiratory support while 11.5% required invasive support

COVID-19 mortality

• Overall mortality rate from COVID-19 was 28.6% (50 patients) after a median of 9.5 days after diagnosis, which did not significantly differ between participating countries. This was higher than the general population based on the case-fatality rate at the time in the general population in Spain (23%), and in Italy (12.8%)

• Patients in ICU had the worst prognosis with a mortality rate of 50% compared with 25% for patients treated on general wards, and 6% for patients treated at home
• The mortality was higher in COVID-19 patients with MF compared to patients with other types of MPN (48.0% versus ~25%; p = 0.016)
• Similar to the general population, lower survival rates were also seen in patients with chronic heart failure, diabetes, older age, low oxygen saturation, need of hospitalization, noninvasive and invasive respiratory support, and those being transferred to ICU

MPN therapy during COVID-19

• At the time of their COVID-19 diagnosis, 80% of the patients were on MPN-directed therapy, in particular, hydroxyurea (HU; 45.1%), followed by ruxolitinib (25.7%), interferon (2.3%), anagrelide (4.6%), and other drugs (2.3%)
• Treatment discontinuation after COVID-19 diagnosis was seen in 17.3% of patients on HU, and in 24.4% of patients on ruxolitinib
• Discontinuing ruxolitinib led to far worse outcomes than continuing treatment, with a survival probability at 60 days post-COVID-19 diagnosis of 11% in patients who discontinued versus 68% in patients who continued ruxolitinib or 75% in patients who did not receive ruxolitinib

Risk factors for mortality after multivariable analysis

• The most relevant risk factors for survival in multivariable analysis were age, male gender, admission to ICU, need for respiratory support, and ruxolitinib discontinuation (see Table 1)
• The negative effect of ruxolitinib discontinuation was confirmed after propensity matching of 44 of the ruxolitinib-treated patients for factors linked to MPN severity and COVID-19
• The reasons for discontinuation of treatment, such as uncertainty over adverse events, and interactions with other drugs, were not collected and could lead to a potential bias

Table 1. Multivariable analysis of risk factors for mortality in patients with COVID-19¹

Conclusion

Mortality in patients with MF after a COVID-19 diagnosis was higher than in patients with ET, PV, or pre-PMF. None of the MPN-directed therapies were associated with a higher risk of death when used in patient diagnosed with COVID-19.

However, rapid ruxolitinib discontinuation after COVID diagnosis was associated with a significantly higher mortality risk, as shown by multivariable analysis. A possible explanation could be the induction of proinflammatory cytokine release after ruxolitinib discontinuation which may aggravate COVID-19.

The authors acknowledge that their subgroup of 11 patients who discontinued ruxolitinib treatment is small, but the analysis was adjusted for multiple confounders and remained robust after a sensitivity analysis. They suggest that their observation should be spread among patient and clinician communities while a larger observational study with longer follow-up can confirm the results.