Effectiveness of BA.1-and BA.4/BA. 5-Containing Bivalent COVID-19 mRNA Vaccines Against Symptomatic SARS-CoV-2 Infection During the BA.5-Dominant Period in Japan

mRNA COVID-19 vaccines initially showed high efficacy and effectiveness, but waning immunity and the emergence of immune-evading variants raised concerns. To address this, bivalent vaccines containing mRNA for the ancestral strain plus either Omicron BA.1 or BA.4/BA.5 were developed and approved in Japan on September 20 and October 13, 2022, respectively. As approvals were based on in vitro and animal data, robust real-world effectiveness estimates were needed. Japan, with over two thirds of the population infection-naive as of mid-November 2022 and a stable testing strategy, provided a suitable setting to evaluate real-world vaccine effectiveness against symptomatic SARS-CoV-2 infection during the BA.5-dominant period. This study aimed to estimate the absolute and relative effectiveness of BA.1- and BA.4/BA.5-containing bivalent mRNA vaccines against symptomatic infection.

The study references prior observations of high initial vaccine effectiveness for monovalent mRNA vaccines and subsequent waning in the context of emerging variants with immune escape. It notes limited published effectiveness data for BA.4/BA.5-containing bivalent vaccines, largely focused on severe outcomes, and contrasts with a US study reporting lower effectiveness against symptomatic infection. The authors also reference evidence and hypotheses regarding immune imprinting potentially affecting responses to updated vaccines.

Design: Multicenter, prospective, test-negative, case-control study (FASCINATE study). Setting and period: Ten outpatient healthcare facilities in the Kanto region of Japan (Tokyo and three surrounding prefectures), September 20 to December 31, 2022, when BA.5 accounted for an estimated 75%–100% of infections. Participants: Individuals aged ≥16 years presenting with COVID-19-like symptoms. Exclusions: lack of consent, immediate lifesaving treatment needed, prior study participation; for analysis, excluded those with unknown symptom onset time, testing ≥15 days after symptom onset, receipt of non-mRNA or unknown vaccine types. Case/control definition: SARS-CoV-2 PCR performed at participating facilities or commercial labs for diagnosis; PCR-positive were cases, PCR-negative controls. Exposure classification: Vaccination status categorized into 17 groups: (1) unvaccinated; (2) dose 1 or ≤13 days after dose 2; (3) 14–90 days after dose 2; (4) 91–180 days after dose 2; (5) >180 days after dose 2; (6) ≤13 days after dose 3; (7) 14–90 days after dose 3; (8) 91–180 days after dose 3; (9) >180 days after dose 3; (10) ≤13 days after dose 4; (11) 14–90 days after dose 4; (12) 91–180 days after dose 4; (13) >180 days after dose 4; (14) ≤13 days after BA.1-containing bivalent; (15) ≥14 days after BA.1-containing bivalent; (16) ≤13 days after BA.4/BA.5-containing bivalent; (17) ≥14 days after BA.4/BA.5-containing bivalent. Vaccination data were collected via questionnaire prior to test results, corroborated with vaccine records/certificates and plausibility checks. Outcomes: Symptomatic SARS-CoV-2 infection (PCR-confirmed). Covariates: Age group, sex, comorbidity, occupation (healthcare/long-term care worker or not), SARS-CoV-2 diagnostic test in the past month, self-reported past SARS-CoV-2 infection (by period), history of close contact, healthcare facility, calendar week, mask-wearing, high-risk behavior (dining at night with alcohol in a group), and influenza vaccination status (2022–2023 season). Statistical analysis: Logistic regression estimated adjusted odds ratios (aORs) of vaccination among cases vs controls. Vaccine effectiveness (VE) computed as (1 − aOR) × 100%. Absolute VE (aVE) compared vaccinated to unvaccinated. Relative VE (rVE) compared bivalent recipients to those who had received monovalent vaccines 3–6 months earlier or >6 months earlier, regardless of number of monovalent doses. A head-to-head comparison of ≥14 days after bivalent vs 14 days–3 months after the 3rd or 4th monovalent dose was conducted. Influenza vaccination status was examined as a bias indicator; influenza activity was very low during the study period. Analyses used STATA 17.0. Ethics approval obtained with waiver of informed consent.

Participants: 6,955 enrolled; excluded 170 (unknown onset date), 33 (tested ≥15 days after onset), and additional exclusions, yielding 6,191 analyzed (3,498 test-positive; 2,693 test-negative) across 10 facilities. Descriptive characteristics (selected): Median days from onset to test: 1 (IQR 1–2). Comorbidity present in 24.6%. Prior SARS-CoV-2 infection reported by 10.6% overall (2.7% of cases vs 20.8% of controls). Vaccination status: 10.9% unvaccinated; most had received 3 doses (48.1%). Bivalent recipients: BA.1 n=227 (overall), BA.4/BA.5 n=344 (overall). Absolute VE against symptomatic infection (vs unvaccinated): - ≥14 days after BA.1-containing bivalent: aOR 0.35 (95% CI 0.23–0.53), VE 65% (95% CI 47–77). - ≥14 days after BA.4/BA.5-containing bivalent: aOR 0.24 (0.17–0.35), VE 76% (65–83). - ≤13 days after BA.1-containing bivalent: VE 71% (49–86); ≤13 days after BA.4/BA.5-containing bivalent: VE 68% (49–80). Selected monovalent aVE (vs unvaccinated): - 14–90 days after dose 3: aOR 0.24 (0.16–0.35), VE 76% (65–84). - 91–180 days after dose 3: aOR 0.45 (0.35–0.58), VE 55% (42–65). - >180 days after dose 3: aOR 0.50 (0.40–0.63), VE 50% (37–60). - 14–90 days after dose 4: aOR 0.33 (0.23–0.47), VE 67% (53–77); 91–180 days after dose 4: aOR 0.39 (0.26–0.59), VE 61% (41–74). Relative VE of bivalent vs monovalent (regardless of dose number): - Versus 3–6 months after monovalent dose: ≥14 days after bivalent aOR 0.65 (0.49–0.85), rVE 35% (15–51); ≤13 days after bivalent aOR 0.72 (0.49–1.04), rVE 28% (−4 to 51). - Versus >6 months after monovalent dose: ≥14 days after bivalent aOR 0.54 (0.42–0.70), rVE 46% (30–58); ≤13 days after bivalent aOR 0.60 (0.42–0.86), rVE 40% (14–58). Head-to-head comparison: Bivalent ≥14 days vs 14 days–3 months after 3rd/4th monovalent dose: aOR 0.99 (0.72–1.36), VE 1% (−36 to 28), indicating no superiority of bivalent shortly after vaccination. Behavioral and bias assessment: No association between influenza vaccination and SARS-CoV-2 testing was observed (as a bias check).

Bivalent BA.1 and BA.4/BA.5 mRNA vaccines demonstrated high absolute effectiveness against symptomatic BA.5-period infections compared with no vaccination, and moderate additional protection compared with individuals who received monovalent vaccination months earlier. Compared with US data, the aVE estimates against symptomatic infection were higher in this Japanese cohort, potentially reflecting lower prior infection prevalence and stringent public health measures (such as widespread mask use) that may reduce residual confounding. Relative effectiveness was greater when the comparison group had a longer interval since last monovalent dose, consistent with waning of monovalent protection over time. In a head-to-head comparison with recent monovalent boosting (14 days–3 months after dose 3 or 4), bivalent vaccination did not show superiority during the BA.5-dominant period. Overall, while aVE was robust versus unvaccinated groups, it was lower than the historically higher effectiveness observed for monovalent primary series against earlier variants (ancestral, Alpha, Delta). The pattern aligns with the concept of immune imprinting potentially influencing responses to updated vaccines. These findings support the added benefit of bivalent boosters, particularly as time since prior monovalent vaccination increases, while highlighting the complex immunologic landscape affecting incremental gains over recent monovalent boosting.

Bivalent COVID-19 mRNA vaccines (BA.1 and BA.4/BA.5 formulations) provided high protection against symptomatic SARS-CoV-2 infection compared with no vaccination and offered moderate additional protection compared with monovalent vaccines administered more than six months earlier. Although there were indications of immune imprinting and no clear superiority over recent monovalent boosting, the results support continued rollout of bivalent vaccines. Ongoing monitoring is needed to assess mid- to long-term effectiveness.

Potential biases and confounding inherent to observational designs remain despite adjustment for numerous covariates. Vaccination histories could not be linked to registries; instead, participants referenced vaccine records/certificates or personal diaries/calendars, introducing potential misclassification. Some estimates had wide confidence intervals, warranting cautious interpretation. Analyses were complete-case only. Effectiveness estimates were short term, limited to the BA.5-dominant period, and longer-term durability was not assessed.