From September 26 to 28, 2020, 108 volunteers, who received an Ad5-nCoV prime dose 6 months earlier in the previous phase 1 trial, were recruited and screened for enrollment (Fig. S1). A total of 89 eligible participants, including low prime dose group (5 × 1010 vp, n = 28), middle prime dose group (1 × 1011 vp, n = 30), and high prime dose group (1.5 × 1011 vp, n = 31), were enrolled. The mean age of the participants was 38.0 years (SD 10.8), with 82 (92%) individuals aged 18–55 years and 7 (8%) aged 56–60 years (Table 1). Forty-six (52%) of the 89 participants were male. Baseline characteristics did not differ considerably across the prime dose groups. The proportion of participants with high pre-existing Ad5 immunity was 64% (18/28) in the low prime dose group, followed by 50% (15/30) in the middle prime dose group, and 45% (14/31) in the high prime dose group, but there was no significant difference among the three prime dose groups (p = 0.316). All the participants received the booster dose and completed the scheduled safety observation within 28 days post vaccination. Eighty-eight (99%) individuals completed blood sample collection at day 14 and 28.

Seventy-two (81%) of the 89 participants reported at least one adverse reaction within the first 14 days after booster: 23 (82%) in the low prime dose group, 23 (77%) in the middle prime dose group, and 26 (84%) in the high prime dose group (Table 2). No significant difference in the number of adverse reactions within 14 days (p > 0.05) across the three prime groups was observed, nor in the overall number of adverse events within 28 days (p = 0.837). In all groups, most adverse reactions (> 60% grade 1) were mild or moderate: pain at the injection site and fatigue were the most common adverse reactions. Injection site induration in the individuals with prior high dose was the most prevalent (p = 0.008). No clinically significant abnormal changes in hematological indexes were observed at day 1. Two pregnancy events occurred within 12 months after booster, and no abnormal pregnancy outcome was observed. Besides, one severe adverse event was reported but not considered vaccine related, which was hospitalization for erysipelas in the right lower limb over 8 months after the booster dose, and this participant finally recovered after symptomatic treatment.

Specific binding antibody responses to RBD were determined via ELISA (Table 3). At day 28, the participants in the prior low dose group had a higher GMT (1124.7, 95% CI 762.8–1658.2) of binding antibody, compared with 543.5 (384.5–768.1) for the middle and 639.0 (476.1–857.6) for the high prime dose group. Meanwhile, 28 (100%) individuals in the prior low dose group, 28 (97%) in the prior middle dose group, and 30 (97%) in the high prime dose group had at least a fourfold increase (seroconversion) in binding antibody titers at day 28. No significant difference in seroconversion rate (p > 0.999) was shown across the three groups. Neutralizing antibodies against live SARS-CoV-2 had a moderate increase at day 14 compared to baseline and peaked at 28 days after booster. The GMTs of neutralizing antibody were noted as 101.8 (65.7–157.7), 42.5 (30.2–59.8), and 58.9 (41.3–84.0) in the low, middle, and high prime dose group at day 28, respectively. Eighty (91%) of 88 individuals experienced at least a fourfold increase of neutralizing antibody titers: all 28 recipients (100%) in the low prime dose group, 24 (83%) of 29 in the middle prime dose group, and 28 (90%) of 31 in the high prime dose group on day 28. For neutralizing antibodies against pseudovirus, all three groups had 100% seroconversion both at days 14 and 28. ELISA binding antibodies to RBD showed positive correlation of 0.766 with neutralizing antibodies against live virus and the association between binding antibodies and neutralizing antibodies to pseudovirus was 0.845 at day 28 (Fig. S2).

Interestingly, all participants (100%) with low pre-existing Ad5 immunity prior to vaccination had a fourfold increase in neutralizing antibody titers at 28 days after booster, whereas 18 individuals (100%) in the low prime dose group, 9 (64%) of 14 in the middle prime dose group, and 11 (79%) of 14 in the high prime dose group, who had high pre-existing Ad5 immunity prior to vaccination, had at least a fourfold increase in neutralizing antibody titers at 28 days after booster. This showed that high pre-existing anti-Ad5 vector immunity moderately impaired the effectiveness of antibody responses, albeit to a fairly acceptable level (Table S1). Moreover, the levels of Ad5 neutralizing antibodies significantly improved post booster (Table S2).

IFNγ of specific T cell responses was determined with ELISpot as the number of SFCs per 105 cells. The proportions of ELISpot-positive responses ranged from 86% to 94% across the three prime dose groups at 14 days post booster with a median number of SFCs per 105 cells of 17.5 (interquartile range (IQR) 10.0–29.8) in the prior low dose group, 15.0 (7.0–21.0) in the prior middle dose group, and 14.0 (6.0–22.0) in the prior high dose group (Fig. 1). No significant difference was shown across the three groups (p = 0.604). Compared to those with low Ad5 neutralizing antibodies at baseline, high pre-existing Ad5 immunity slightly reduced the proportion of ELISpot-positive responders after booster: 16 (89%) of 18 participants in the low prime dose group, 11 (79%) of 14 in the middle prime dose group, and 13 (93%) of 14 in the high prime dose group, who had high Ad5 neutralizing antibody titers prior to vaccination, were identified as positive responders at day 14.

IFNγ of specific T cell responses measured by ELISpot. a Number of IFNγ-secreting cells per 105 PBMCs at day 0 and 14 in all participants and stratified by the pre-existing anti-Ad5 neutralizing antibody titers. b Proportion of positive responders at day 14 post booster in all participants and stratified by the pre-existing anti-Ad5 neutralizing antibody titers. Error bars are interquartile ranges (IQRs). ELISpot enzyme-linked immunospot assay, IFNγ interferon gamma, PBMCs peripheral blood mononuclear cells, Ad5 adenovirus type 5

IFNγ and TNFα expression was detected by ICS. The booster vaccination at day 14 enhanced the specific IFNγ and TNFα responses in CD4+ T cells in all groups, but no significant difference was found with an overall p value of 0.357 for IFNγ responses (0.634 for TNFα) across the three prime dose groups (Fig. 2). Similar CD8+ T cell responses were observed. The further stratified analysis showed that the pre-existing anti-Ad5 neutralizing antibody had an acceptable effect on IFNγ and TNFα responses in CD4+ and CD8+ T cells (Fig. S3).

Spike protein-specific IFNγ and TNFα responses in CD4+ and CD8+ T cells detected by ICS. a Percentage of CD4+ T cells producing IFNγ and TNFα at day 0 and 14. b Percentage of CD8+ T cells producing IFNγ and TNFα at day 0 and 14. Error bars are interquartile ranges (IQRs). ICS intracellular cytokine staining, IFNγ interferon gamma, TNFα tumor necrosis factor alpha

Besides, because COVID-19 cases were sporadic in China at that time, no natural SARS-CoV-2 infection was found during the study period.