The impact of Covid-19 in patients with chronic myeloid leukemia—a nationwide population-based study

were investigated: the risk of testing positive for Covid-19, hospital or ICU admission, respectively, related to Covid-19 (de ﬁ ned as hospital or ICU admission with Covid-19 as a primary or secondary diagnosis within 60 days of the ﬁ rst positive Covid-19-test) or death (de ﬁ ned as death with Covid-19 as underlying or contributing cause of death during follow-up). In several secondary analyses

We defined two separate CML cohorts. The first, main cohort, consisted of individuals diagnosed with TKI-treated CML in chronic phase (CP-CML) before March 1st 2020, defined here as the start of the pandemic in Sweden. A secondary cohort consisted of individuals diagnosed during the pandemic, after this index date. We excluded individuals born outside of Europe, as it is documented that they suffered increased susceptibility to severe Covid-19 outcomes in the Swedish population [9]. Using the same database, we selected five controls for each CML patient in the two cohorts by randomly matching for age (within 60 days), sex, and county of residence on March 1st 2020. The individuals in the two CML and control cohorts were followed from March 1st 2020 to the first event of each outcome studied, April 1st 2022, emigration or death, whichever occurred first. This effectively means that follow-up for patients diagnosed during the pandemic was also initiated at the start of the pandemic and not at the time of the CML diagnosisallowing to capture potentially more severe Covid-19 cases even if diagnosed before the CML diseasea disease that is diagnosed in the asymptomatic case in half of the patients and that has existed and affected the body for years before the diagnosis date.
We performed multiple sets of analyses, comparing the incidence rate ratios (IRRs) between the CML and control cohorts. In the first main analysis, elucidating the inherent risk of Covid-19 in an unvaccinated CML population, patients were additionally censored at the time of first vaccination. Four separate outcomes were investigated: the risk of testing positive for Covid-19, hospital or ICU admission, respectively, related to Covid-19 (defined as hospital or ICU admission with Covid-19 as a primary or secondary diagnosis within 60 days of the first positive Covid-19-test) or death (defined as death with Covid-19 as underlying or contributing cause of death during follow-up).
In several secondary analyses, we investigated the same four outcomes. First by initiating follow-up at the time of vaccine protection (defined as 14 days after the 2nd dose of any vaccine used), comparing the CML cohort to their control cohort. To reduce the impact of the knowledge gap regarding the management of Covid-19 disease early in the pandemic and the potential for less adequate management, we also performed analyses limiting the population to CML patients and their respective controls who were alive and without any of the outcomes after the first 6 months of the pandemic. We also compared the cohort consisting of patients diagnosed during the pandemic to their matched control cohort to elucidate the impact of a more active, recent leukemic disease. Furthermore, we investigated the effect of the first-line TKI imatinib as compared to the later generations TKIs (dasatinib, nilotinib, bosutinib, or ponatinib) at the time of Covid-19-infection, limiting the analysis to the CML population. IRRs were calculated using a quasi-Poisson regression model incorporating sex (categorical) and age (as a restricted cubic spline with 3 equally placed knots), in the main analysis to model the events comparing the CML cohort with the matched control cohort. In the analysis within the CML cohort, the model incorporated the same covariates with the addition of TKI treatment as a categorical time-dependent variable (imatinib or any other TKI). A total of 1146 CP-CML patients were identified and 5,730 matched controls were selected. Among them, 83% of the CML patients were diagnosed before the index date March 1st 2020. The characteristics of patients and controls are outlined in Table 1. The median follow-up from the index date was 1.3 years for all groups.
The risk of Covid-19 infection and severe Covid-19 disease among CML patients is summarized in Table 2. The risk of testing positive for Covid-19 was not significantly increased in CML patients. Among unvaccinated CML patients, 106 (11.1%) developed positivity, as compared to 448 (9.5%) among controls.
However, CML patients in the main cohort (i.e., diagnosed prior to the index date) and with no Covid-19-vaccination showed an increased risk of hospitalization for Covid-19, as compared to controls; IRR 1.93 (95% confidence interval [CI], 1.17-3.18, P = 0.01). Among the 24 unvaccinated CML patients admitted to the hospital due to Covid-19, all had comorbid conditions. The most common comorbid conditions were cardiovascular diseases (N = 15), renal disease (N = 12), diabetes (N = 10), and chronic lung disease (N = 8). A similar point estimate, but due to lower statistical power, non-significant trend was observed for mortality with IRRs for Covid-19-related death 2.04 (95% CI, 0.91-4.56). This corresponds to a case fatality-rate of 1.9% (N = 2) and 1.1% (N = 5) for CML patients and controls, respectively. None of the CML patients who tested Covid-19 positive were admitted to an ICU.
In the analysis where we delayed entry with 6 months, to account for incomplete testing and a high mortality rate early in the pandemic phase, the effect of increased Covid-19 hospital admission disappeared.
The vaccination rate during the follow-up was high among both CML patients and controls; 94% and 92%, respectively. Initiating follow-up at the time of vaccination for each subject, there was no significant difference between the CML and the control cohorts regarding any of the outcomes, likely due partly to fewer events, although point estimates were consistent with the ones for unvaccinated individuals.
Focusing on the 198 CML patients diagnosed after the index date March 1st 2020 (i.e., during the pandemic) we identified no increased risk vs. their 948 matched controls for any of the listed outcomes Studying the effect of imatinib vs. later generation TKIs we detected no significant differences related to any of the Covid-19 outcomes. In a sensitivity analyses relaxing the definition of hospital admission and/or ICU admission to not require a previous positive PCR test, we did not see any altered findings.
In summary, in this register-based matched cohort study utilizing high-quality nationwide data from a large number of individuals, we investigated the risk of developing Covid-19 and suffering more severe Covid-19 outcomes between CML patients and carefully matched population controls. The overall risks for developing Covid-19 positivity, ICU admission, and case-related mortality were comparable and relatively low. However, in CML patients with no Covid-19 vaccination, we observed an increased risk of hospitalization, as compared to controls. In this study, however, we cannot accurately describe the association between CML and Covid-19 death due to the low number of deaths that occurred. Still, the few deaths observed suggest that if there is an increased risk of death, this increased risk is moderate.
We did not observe any distinct differences regarding Covid-19 outcome comparing CML patients on 1st generation TKI (imatinib) vs. later generations TKI. Moreover, patients diagnosed with CML after the onset of the Covid-19 pandemic (i.e., after the index date March 1st 2020), and thus with a higher leukemic burden, did not display any increased risk of suffering more severe outcomes.

DATA AVAILABILITY
Due to laws guarding the integrity of Swedish citizens, no patient-level data can be made available. Upon request, aggregate data with different levels of resolution may be possible to provide.    ICU intensive care unit. *2G TKI included dasatinib, nilotinib, bosutinib and ponatinib.

AUTHOR CONTRIBUTIONS
T. Dahlén et al.