Infusion reactions in natural killer cell immunotherapy: a retrospective review

Background aims: The use of natural killer (NK) cells as a cellular immunotherapy has increased over the past decade, specifically in patients with hematologic malignancies. NK cells have been used at the authors’ institution for over 15 years. Most patients have a reaction to NK cell infusion. The authors retrospectively analyzed the reactions associated with NK cell infusions to characterize the types of reactions and investigate why some patients have higher-grade reactions than others. Methods: A retrospective chart review of NK cell infusions was performed at the authors’ institution under nine clinical protocols from 2008 to 2016. An infusion reaction was defined as any symptom from the time of NK cell infusion up to 4 h after infusion completion. The severity of infusion reactions was graded based on Common Terminology Criteria for Adverse Events, version 4. Two major endpoints of interest were (i) infusion reaction with any symptom and (ii) grade ≥3 infusion reaction. Multivariable logistic regression models were used to investigate the association between variables of interest and outcomes. Odds ratios (ORs) and 95% confidence intervals (CIs) were obtained for each variable. Results: A total of 130 patients were receiving NK cell infusions at the authors’ institution. The most common reported symptom was chills (n = 110, 85%), which were mostly grade 1 and 2, with only half of patients requiring intervention. There were 118 (91%) patients with infusion reactions, and only 36 (28%) were grade 3. There was one life-threatening grade 4 reaction, and no death was reported due to infusion reaction. Among grade ≥3 reactions, cardiovascular reactions (mainly hypertension) were the most common, and less than half of those with hypertension required intervention. NK cell dose was not associated with any of the grade 3 infusion reactions, whereas monocyte dose was associated with headache (grade ≤3, OR, 2.17, 95% CI, 1.19–3.97) and cardiovascular reaction (grade ≥3, OR, 2.13, 95% CI, 1.13–3.99). Cardiovascular reaction (grade ≥3) was also associated with in vitro IL-2 incubation and storage time. Additionally, there was no association between grade ≥3 infusion reactions and overall response rate (OR, 0.75, 95% CI, 0.29–1.95). Conclusions: The majority of patients who receive NK cell therapy experience grade 1 or 2 infusion reactions. Some patients experience grade 3 reactions, which are mainly cardiovascular, suggesting that close monitoring within the first 4 h is beneficial. The association of monocytes with NK cell infusion reaction relates to toxicities seen in adoptive T-cell therapy and needs further exploration.