Using machine learning to predict antibody response to SARS-CoV-2 vaccination in solid organ transplant recipients: the multicentre ORCHESTRA cohort

Abstract

Objectives: The study aim was to assess predictors of negative antibody response (AbR) in solid organ transplant (SOT) recipients after the first booster of SARS-CoV-2 vaccination.

Methods: Solid organ transplant recipients receiving SARS-CoV-2 vaccination were prospectively enrolled (March 2021-January 2022) at six hospitals in Italy and Spain. AbR was assessed at first dose (t₀), second dose (t₁), 3 ± 1 month (t₂), and 1 month after third dose (t₃). Negative AbR at t₃ was defined as an anti-receptor binding domain titre <45 BAU/mL. Machine learning models were developed to predict the individual risk of negative (vs. positive) AbR using age, type of transplant, time between transplant and vaccination, immunosuppressive drugs, type of vaccine, and graft function as covariates, subsequently assessed using a validation cohort.

Results: Overall, 1615 SOT recipients (1072 [66.3%] males; mean age±standard deviation [SD], 57.85 ± 13.77) were enrolled, and 1211 received three vaccination doses. Negative AbR rate decreased from 93.66% (886/946) to 21.90% (202/923) from t₀ to t₃. Univariate analysis showed that older patients (mean age, 60.21 ± 11.51 vs. 58.11 ± 13.08), anti-metabolites (57.9% vs. 35.1%), steroids (52.9% vs. 38.5%), recent transplantation (<3 years) (17.8% vs. 2.3%), and kidney, heart, or lung compared with liver transplantation (25%, 31.8%, 30.4% vs. 5.5%) had a higher likelihood of negative AbR. Machine learning (ML) algorithms showing best prediction performance were logistic regression (precision-recall curve-PRAUC mean 0.37 [95%CI 0.36-0.39]) and k-Nearest Neighbours (PRAUC 0.36 [0.35-0.37]).

Discussion: Almost a quarter of SOT recipients showed negative AbR after first booster dosage. Unfortunately, clinical information cannot efficiently predict negative AbR even with ML algorithms.

Keywords: Antibody response; COVID-19; Machine learning; SARS-CoV-2; Solid organ transplantation; Vaccination.

Fig. 1

Distribution of antibody response. Individuals are classified as having a no antibody response if their antibody level is between 0 and 5.58 Bau/ml, Inconclusive if the level is between 5.58 and 45, positive-low the level is between 45 and 205, positive-mild if the level is between 205 and 817 BAU/ml, and classified as having a high antibody response if their antibody level is above 817 BAU/ml. Transitions between bars show the transition fractions of individuals across time points.

Fig. 2

Parameter estimates ordinal logistic regression. Results of the ordinal logistic regression model using the depicted covariates and the 5 categories “None”, “Inconclusive”, “Positive-low”, “Positive-mild”, and “High” with None being encoded as the state with the lowest antibody response and High being the state with the highest antibody response. Hence, negative coefficients indicate a more negative antibody response. Confidence intervals are at the 95% level. The coefficients can be interpreted as an increase in the log odds ratio, if the respective control variable increases by one. There are only four out of the six centers included in the graph since one center had only observations with missing data at the 3rd vaccination and one center has no parameter since it is the reference group. The confidence intervals for age are due to their size not properly depicted in the graph. However, its 95% confidence interval does not cover zero. Exact values and p-values are given in Supplemental Table 6. The transplant results are in comparison to liver transplants, the timing of vaccination in comparison to less than one year and the Pfizer vaccine parameters in comparison to Moderna.