Background: High dose chemotherapy (HDT) burdens patients with a high risk of serious infections while neutropenic. The role of host antibodies as risk factors for infections is unclear. Our aim was to investigate if pre-HDT levels of IgG antibodies against terminal Galα3Gal (anti-αGal) predict infections. Anti-αGal is reported to be the most abundant antibody in human plasma and is able to recognize most sepsis-causing Gram-negative bacteria. The antibody exerts poor complement activation, which promotes pathogen survival by blocking access of more effective complement activators. This may be particularly problematic in neutropenic patients. Thus, we hypothesized that patients with high anti-αGal serum concentrations are particularly prone to suffer infections following HDT.

Methods: We conducted a clinical cohort study on patients ≥16 years receiving HDT with autologous stem cell transplantation for myelomatosis (MM) or non-Hodgkin lymphoma (NHL) at Department of Hematology, Aarhus University Hospital, Denmark from 25 November 2009 through 26 June 2015. Of eligible patients (N=308), we excluded previous transplanted (N=14), those without a pre-HDT serum sample available for anti-αGal quantification (N=115), and recipients of plasma transfusion/IgG substitutions within 90 days before the pre-HDT serum sample and until end of follow-up (N=9). All included received prophylactic antimicrobial therapy (levofloxacin and fluconazol) and filgrastim. Serum anti-αGal was quantified using an in-house Time-resolved Immuno-Fluoremetric Assay. We ascertained infectious episodes within 30 days following the date of autologous stem cell re-infusion through review of all medical charts and data retrieval from the laboratory information systems, blinded for anti-αGal concentrations. Infectious episodes were defined as fever (≥38.5°C) or hypothermia (<36°C) prompting antibiotic treatment and CRP > 21 mg/L preceded by at least 24 hours without any of these. We used a Cox proportional hazards model to investigate the association between anti-αGal concentrations and risk of infection, with adjustment for total plasma IgG concentrations, ABO blood group, comorbidity, and underlying disease (MM or NHL) in the adjusted hazard ratios (HR). Optimum cutoff for dichotomizing was determined from ROC analysis applying Youden´s J statistic and groups were compared by Kaplan-Meier method.

Results: We included 170 patients (MM, 55%; males, 58%; median age 62 years). Severe neutropenia was transient in all, with blood neutrophils increasing to above 500/µL on median day 11 (range 8-18). We identified one infectious episode in 103 patients (61%) and two episodes in six (3.5%) patients. The infectious episodes began during severe neutropenia in 92% of the patients. The infection types were fever of unknown origin (FUO) (72%), pneumonia (19%), catheter related (4.3%), typhlitis (1.7%), C. difficile colitis (1.7%), and hypothermia of unknown origin (0.87%). None died from infections. Pre-HDT serum was collected on median day -31 (range -142- -16). Anti-αGal was detectable in all but one sample, averaging 0.51 mg/dL (range: 0.016-12). Overall, anti-αGal concentrations predicted infectious episodes (all types), crude HR 1.2 (95% confidence interval (CI) 1.1-1.4), adjusted HR 1.1 (95% CI: 1.0-1.3). Using anti-αGal at 1.3 mg/dL as cut-off, we identified that patients with higher concentrations (N=41) experienced considerably increased risk of infectious episodes (crude estimates): all types, HR 1.7 (95% CI: 1.3-3.2); FUO, HR 2.0 (95%CI: 1.5-4.5); and pneumonia HR 2.1 (95%CI: 0.83-8.7).

Discussion: We found that high anti-αGal serum concentrations predict an increased risk of infection after HDT. These findings support that anti-αGal may critically enhance pathogen-survival in the neutropenic host. Studies of other cohorts and mechanistic studies are required. Our findings may pave the way for future optimized risk stratification and therapeutic measures.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.