Comprehensive viromewide antibody responses by systematic epitope scanning after hematopoietic cell transplantation

Rachel A. Bender Ignacio, Sayan Dasgupta, Terry Stevens-Ayers, Tomasz Kula, Joshua A. Hill, Stephanie J. Lee, Marco Mielcarek, Ann Duerr, Stephen J. Elledge and Michael Boeckh

Data supplements

Article Figures & Data


  • Figure 1.

    Visual representations of VirScan scores over time in HCT pairs. (A) Aggregated heat map with 5 longitudinal samples, grouped by CMV serostatus of the HCT donor-recipient pair (D+/R+, D/R+, D+/R, D/R). Colors correspond to mean numbers of epitopes recognized within all samples belonging to that group (range <1 to 40 mean recognized epitopes). White designates no epitope recognition in any sample (mean = 0). Attention to CMV row (red label) shows distinct differences over time by CMV serostatus as determined by Food and Drug Administration–approved serology. A threshold of 6 VirScan epitopes performs optimally against clinical serology (B) CMV epitopes detected by VirScan vs clinical reactivation of CMV in a D/R+ recipient who also received high daily doses of systemic steroids for GVHD. Each quantitative antigenemia value is depicted by a purple dot and daily steroid dose by the orange line. (C) Acquisition of respiratory syncytial virus (RSV) epitopes during lower respiratory tract infection in a susceptible patient (D/R+ for RSV); RSV was detected in upper and lower respiratory samples by direct fluorescent antibody (DFA) on 4 individual samples (DFA in clinical use, shown with asterisks). HPV, human papillomavirus; MPV, human metapneumovirus; HHV, human herpesvirus; SARS, severe acute respiratory syndrome; pre-D, pre-HCT donor sample; pre-R, pre-HCT recipient sample.

  • Figure 2.

    Pairwise diversity comparison between recipient and donor antiviral repertoires from pre-HCT through 1 year following transplant. Higher β-distance represents greater dissimilarity between the repertoire of the donor and recipient at any time point (n = 37). Recipient repertoires become more similar to the donor until d100 (β = 0.381) and then once again diverge by d365 (β = 0.407); pre-HCT recipient and d365 comparisons with donor were not different. The mean β-distance between recipient pre-HCT samples and d365 samples was 0.321, meaning that 1 year after HCT, the antiviral repertoire is significantly more similar to pre-HCT self than that of the donor (P < .001; see text), as modeled with an LME model. pre-tx, pretransplant.

  • Figure 3.

    Donor and recipient age predicts antiviral antibody repertoires. Overall, persons in middle age (40-55 years) had higher epitopes scores, total viruses recognized, and higher diversity compared with both younger and older donors and recipients, and donor age more highly predicted d100 scores (n = 11, <40 years; n = 15, 40-55 years; n = 11, >55 years), whereas recipient age was more strongly predictive at d365 (n = 10, <40 years; n = 18, 40-55; n = 19, >55 years). Dots shown are estimates from generalized linear models, and whiskers depict 95% CIs for each estimate. (A) Simpson’s diversity within donor sample vs donor age: a lower score equates to higher levels of diversity of antiviral antibodies. (B) Donor age predicts total recognized epitopes at d100. (C) Donor age predicts all non-CMV epitopes (CMV epitopes removed to adjust for D/R CMV status) also at d100. (D) Recipient age predicts non-CMV epitopes pre-HCT. (E) Recipient age predicts total epitopes at d365. (F) Trend toward recipient age predicting non-CMV epitopes at d365 with significant differences between age 40-55 and >55.

  • Figure 4.

    Donor and recipient CMV status is associated with viral epitope recognition at d100 but not d365 after HCT. (A-C) Dot and whisker plot representing the estimate and 95% CI of total epitopes recognized at d100 (D/R n = 3; D+/R n = 8; D/R+ n = 5; D+/R+ n = 21) (A) and total epitopes excepting CMV epitopes (C), derived from a generalized linear model accounting for both donor and recipient CMV status. D/R+ recognized most epitopes, followed by both D+ groups, which shows that CMV confers greater influence than simply the addition of its epitopes to the total epitope count; there is an interaction between donor-recipient dynamics. (B,D) Similar plot at d365, although CMV status of donor or recipient no longer influences total viral epitopes recognized, whether or not CMV epitopes were included.

  • Figure 5.

    Receipt of high-dose glucocorticoids and GVHD is associated with viral epitope recognition at d100. In univariate analysis, presence of GVHD was not associated with VirScan metrics, but high-dose steroids (>1 mg/kg prednisone equivalents per day given within 2 weeks prior to analysis) was associated with lower total and non-CMV epitopes. Due to interaction of steroid dose and presence of GVHD, we present this analysis stratified by prior diagnosis of GVHD. In the non-GVHD group, <0.5 mg/kg, n = 15; 0.5-1 mg/kg, n = 7; >1 mg/kg, n = 1. In the GVHD group, <0.5 mg/kg, n = 3; 0.5-1 mg/kg, n = 3; >1 mg/kg, n = 1. At d365, there were no statistically significant differences between groups.

  • Figure 6.

    Changes in viral epitope recognition at 1 year post-HCT. HCT pairs were assigned to a D/R status individually for each virus (each HCT pair could contribute to a different category for each virus) based on the described thresholding algorithm. Linear mixed modeling then evaluated group mean gain or loss of epitopes with respect to the pre-HCT sample of the donor (A-D) or recipient (E-H). Viruses for which there was net loss are shown with the estimate and 95% CI in blue (P < .05), and those with net gain are shown in red (P < .05); if the 95% CI crossed 0, black is used. Overall, D/R pairs showed no changes from either D or R baseline (A,E). With respect to D, the D+/R pairs mainly lost epitope recognition across multiple viruses (B) and the D/R+ pairs gained epitopes across many viruses after HCT (C). With respect to either D or R baseline, D+/R+ pairs mainly showed loss of epitopes (D,H). HHVs are shown here alphabetically by number, wherein HHV-1/2 = HSV 1/2; HHV-3 = varicella zoster virus; HHV-4 = Epstein-Barr virus; HHV-5 = cytomegalovirus; HHV6A/B, 7 = roseolaviruses.

  • Figure 7.

    Recognition of individual CMV epitopes among HCT recipients who reactivated or did not reactivate CMV in the first 100 days after hematopoietic cell transplantation. Each row represents a unique CMV peptide, and peptides are ranked in descending order of frequency of recognition among all participants. Each labeled column (eg, P1) contains 2 samples, one from the recipient pre-HCT (pre-R, lighter bars within each color) and the other at 100 days post-HCT (d100, darker bars). Participants with seropositive donors (D+) are on the left, further stratified as to whether they demonstrated CMV reactivation before d100 (≥100 copies of CMV per milliliter by plasma PCR or ≥2 cells positive for CMV pp65/2 × 106 cells), reactivators in blue (n = 15) and nonreactivators in magenta (n = 8). Participants with seronegative donors (D) are also stratified by CMV reactivation (purple; n = 4) or no reactivation (red; n = 4). Visually, those who reactivated (either D+ or D) generally expressed more CMV epitopes. There was a trend toward reactivators having more heterogeneity in expressed epitopes before and after reactivation (42.1% vs 8.3%, P = .1). Three participants (P3, P4, and P25) who had confirmed diagnoses of CMV disease, which was gastrointestinal disease in all cases, are annotated with an asterisk. Six of 37 participants who were missing either sample are not displayed.


  • Table 1.

    HCT cohort demographics and clinical characteristics

    CharacteristicN = 37 pairs (%) or median (IQR)
    Recipient age, y47 (39.5, 54)
    Recipient sex
     Female14 (38)
    Race/ethnicity (donor-recipient)
     White28 (76)
     Asian/Pacific Islander5 (14)
     Hispanic (mixed race)3 (8)
     Black1 (3)
    Donor age48.5 (36, 55)
    Donor type
     Related36 (97)
     Unrelated1 (3)
    Donor sex
     Female22 (59)
     HLA-matched36 (97)
     HLA-mismatched (6/10 match)1 (3)
    Underlying disease
     AML12 (32)
     ALL2 (5)
     CML12 (32)
     CLL2 (5)
     Myelodysplastic syndromes6 (16)
     Non-Hodgkin lymphoma3 (8)
    Disease severity
     Early disease23 (62)
     Intermediate disease7 (19)
     Advanced disease7 (19)
    Cell source*
     Bone marrow11 (30)
     Peripheral blood27 (73)
    Treatment year
     1999-200333 (89)
     2012-20134 (13)
    CMV serostatus of pair
     D+/R+21 (57)
     D+/R8 (22)
     D/R+5 (14)
     D/R3 (8)
    Previous treatment regimens
     Anti-CD203 (8)
     Lymphodepletion13 (35)
     Intensive chemotherapy (any)19 (51)
    Conditioning regimen
     Bu/Cy, + ATG (1) or RAB (1)26 (70)
     TBI, + Cy (5) or VP-16 (1)8 (22)
     Bu, Flu3 (8)
    GVHD prophylaxis
     CSA, MTX31 (84)
     CSA, Cy4 (11)
     CSA, MMF2 (5)
    Systemic GVHD (by NIH criteria)
     Yes25 (68)
    • ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; ATG, antithymocyte globulin; Bu, busulfan; CLL, chronic lymphocytic leukemia; CML, chronic myeloid leukemia; CSA, cyclosporine; Cy, cyclophosphamide; Flu, fludarabine; MMF, mycophenolate mofetil; MTX, methotrexate; NIH, National Institutes of Health; RAB, radiolabeled antibodies; TBI, total body irradiation; VP-16, etoposide.

    • * Totals to 38; 1 person received both types of cells.

    • Anticancer or lymphodepleting agents given within the 12 mo prior to HCT were considered. The only anti-B-cell therapy given in this cohort was rituximab; no ofatumumab or anti-CD52 was given. Regimens considered to be lymphodepleting included rituximab, fludarabine, high-dose cytarabine, or cycloporin/ifosphamide, or combinations of the preceding.

  • Table 2.

    Antiviral repertoire metrics in HCT donors and recipients by time point

    Metric, mean (SD)DonorPre-Rd30d100d365Overall
    Total epitopes123.2 (49.0)124.1 (37.4)136.4 (45.6)139.0 (45.1)111.9 (30.5)126.8 (42.8)
    Non-CMV epitopes106.0 (45.3)106.2 (32.4)116.8 (36.7)118.1 (38.5)91.0 (31.0)107.6 (38.0)
    Total viruses9.4 (3.3)9.5 (3.1)10.7 (3.6)10.5 (2.9)8.5 (3.1)9.7 (3.3)
    Simpson’s D0.10 (0.03)0.11 (0.04)0.10 (0.03)0.10 (0.04)0.12 (0.07)0.10 (0.04)
    Shannon’s evenness0.80 (0.05)0.80 (0.04)0.80 (0.04)0.80 (0.04)0.79 (0.07)0.80 (0.05)