The pandemic of coronavirus disease 2019 (COVID-19) is a serious threat to global health (The Johns Hopkins Coronavirus Resource Center, 2021). It started in Japan with the outbreak in the Diamond Princess Cruise Ship and continued to spread nationwide (Kato et al., 2020). There is hope that durable immunity, namely the so-called immunity passport, would develop after an infection, which is also expected to be achieved with vaccination (Phelan, 2020). Humoral immune responses, especially the production of neutralizing antibodies, play key roles in defending against severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) infection (Stephens and McElrath, 2020). However, there is looming uncertainty regarding the duration and protective efficacy of the humoral immune response following primary infections. Various studies have reported a decline in neutralizing antibodies titers (Crawford et al., 2020; Gaebler et al., 2020; Long et al., 2020; Ripperger et al., 2020; Seow et al., 2020; Wajnberg et al., 2020; Wheatley et al., 2020) and also those of the antibodies against spike protein (SP) or nucleocapsid protein (NP) (Crawford et al., 2020; Gaebler et al., 2020; Ibarrondo et al., 2020; Kutsuna et al., 2020; Long et al., 2020; Patel et al., 2020; Ripperger et al., 2020; Seow et al., 2020; Wajnberg et al., 2020; Wheatley et al., 2020). However, these studies were limited by small sample sizes (Ibarrondo et al., 2020; Kutsuna et al., 2020) and most of the samples in these studies were collected within 2 months after infection (Long et al., 2020; Patel et al., 2020). The decline in antibody titers shown by these studies during the relatively early phase of recovery reflects a reduction in the number of short-lived plasmablasts, which is a natural process of immune response and should not be a concern (Stephens and McElrath, 2020). Subsequent to this decline, long-lived plasma cells in the bone marrow play active roles in maintaining circulating neutralizing antibodies (Stephens and McElrath, 2020). Considering the lacunae of the existing studies, there is a definitive need to determine the extent of duration the neutralizing antibodies persists following natural infection.

A large-scale study in Iceland reported that 1,107 of 1,215 (91.1%) COVID-19 survivors had positive pan-Ig assay results for SP and NP antigens 120 days after diagnosis (Gudbjartsson et al., 2020). However, the study lacked information regarding the NT₅₀ and could not directly determine neutralizing activity persistence. A recent study with 36 samples of COVID-19 survivors at Mount Sinai Health System reported that robust neutralizing antibodies against SARS-CoV-2 infection persisted until 148 days post symptom onset (Wajnberg et al., 2020). However, those samples were selected according to the distribution of available enzyme-linked immunosorbent assay (ELISA) titers; they could therefore not reflect the entire study population. Thus, to the best of our knowledge, no definitive conclusion has been reached regarding the long-term sustainability of neutralizing antibodies after SARS-CoV-2 infection till date. Furthermore, clinical characteristics associated with neutralizing titers in the recovery stage of COVID-19 patients are yet to be understood.

Hence, in this cross-sectional study, we aimed to examine the humoral immune responses, including neutralizing antibodies, at 6 months after SARS-CoV-2 infection among 376 individuals. We studied the NT₅₀ and the titers of antibodies against SP or NP antigens 6 months after the first positive results for SARS-CoV-2, to examine factors associated with NT₅₀, and to investigate the correlation of the NT₅₀ with the other antibody titers.

Between September 2nd and October 26th, 2020, 376 participants with confirmed SARS-CoV-2 infection participated voluntarily in this study. The first date of positive SARS-Cov-2 test result ranged from February 4th to May 26th, 2020 (Supplementary Figure 2). The median (range) age of the participants was 50 (20–78) years, 188 (50%) participants were men, and the median (range) days from the first positive SARS-Cov-2 test result was 180 (147–224) days (Table 1). Among 376 participants, 71 (19%) underwent oxygen support and 25 (6.6%) underwent mechanical ventilation or ECMO during the clinical course. Those who underwent any oxygen support, mechanical ventilation, or ECMO were more likely to be male and older than those who did not. Fever was the most common symptom among 84% of overall participants, 79% had no oxygen support, 96% of participants who underwent oxygen therapy but did not undergo mechanical ventilation or ECMO, and 100% of those who received mechanical ventilation or ECMO had fever. The second common symptom, taste or smell disturbance, was found among 52% of all participants was relatively common among participants who did not undergo oxygen therapy (60%). Cough, the third common symptom, was found among 49% of all participants. Participants that received oxygen or ventilatory support tended to have diabetes.

Overall, most participants tested positive for NT₅₀ (367, 98%), SP-total Ig (369, 98%), SP-IgG (n = 344, 91%), NP-total Ig (n = 365, 97%), NP-IgG (n = 314, 84%), NP-total Ig (n = 365, 97%), and Roche NP-total Ig (n = 367, 98%) (Table 2). In addition, 85% (n = 317) had medium or higher NT₅₀ (≥201). The positive rates of NT₅₀ were 97% among participants without any oxygen support, 100% among those who received oxygen support or required mechanical ventilation or ECMO. Among asymptomatic participants who did not require any oxygen support (N = 14), 11 tested positive for NT₅₀ (79%); among symptomatic participants who did not require any oxygen support (N = 266), 260 tested positive for NT₅₀ (98%) (Supplementary Table 1).

Participants who received oxygen had significantly higher NT₅₀ than those who did not; those that required mechanical ventilation or ECMO had significantly higher NT₅₀ than those who received oxygen alone (Figure 1). We further stratified participants according to five groups according to NT₅₀. Among those without oxygen support, those with higher NT₅₀ tended to have symptoms such as fever and cough, higher BMI, and diabetes; and used favipiravir and steroids (Supplementary Table 2A). Among participants who received oxygen support but did not undergo mechanical ventilation or ECMO, those with higher NT₅₀ tended to have higher BMI (Supplementary Table 2B). Among participants who required mechanical ventilation or ECMO, no factors seemed to be correlated with NT₅₀ (Supplementary Table 2C).

In the multivariable-adjusted model, higher BMI, fever, and the requirement of mechanical ventilation or ECMO were significantly associated with higher titers (Table 3). The NT₅₀ did not differ by days from the first positive results of the SARS-CoV-2 test (Supplementary Figure 3). Spearman’s correlation coefficients between NT₅₀ and antibodies against SP or NP antigens were the strongest for NT₅₀ vs SP IgG (r = 0.63), followed by NT₅₀ vs SP-total Ig (r = 0.57), while antibodies against NP antigen showed weak correlations with NT₅₀ (r = 0.37 for NT₅₀ vs NP IgG; r = 0.32 for NT₅₀ vs NP total Ig; r = 0.33 for NT₅₀ vs Roche NP-total Ig) (Figure 2).

This cross-sectional study of 376 participants across Japan provides evidence that approximately 98% COVID-19 survivors had sustained NT₅₀ (≥51) 6 months after infection. Furthermore, 97% of participants who did not undergo any oxygen support, 100% of those with oxygen support had positive NT₅₀. Those who underwent mechanical ventilation or ECMO were more likely to have higher NT₅₀. In addition, most participants (97 to 98%) tested positive for SP-total Ig, NP-total Ig, and Roche NP-total Ig. These findings provide insight into long-term humoral immune responses after COVID-19 infection, supporting the notion that neutralizing activity remains at detectable levels in most COVID-19 survivors.

With respect to the duration of persistence of neutralizing antibodies, the findings our study are in line with those of certain studies reported previously (Wajnberg et al., 2020; Wu J. et al., 2020). In a study comprising 36 COVID-19 survivors, Wajnberg et al. (2020) have shown that despite their early decline from day 30 until day 82 after infection, NT₅₀ persisted until day 148 and concluded that robust neutralizing antibody responses to SARS-CoV-2 infection persist for months following infection. Seow et al. (2020) have also pointed out the early waning of neutralizing responses to SARS-CoV-2 infection. In this study from the UK, the NT₅₀ were found to wane at day 40 or later following the onset of COVID-19 symptoms and had reduced titers at 94 days after symptom onset (Seow et al., 2020). This decline in neutralizing antibodies at a relatively early phase of infection has been consistently reported in many studies with varieties of populations (Crawford et al., 2020; Gaebler et al., 2020; Ibarrondo et al., 2020; Kutsuna et al., 2020; Long et al., 2020; Patel et al., 2020; Ripperger et al., 2020; Seow et al., 2020; Wajnberg et al., 2020; Wheatley et al., 2020) and is known to occur after the acute phase of infection (Stephens and McElrath, 2020). The crucial question is whether the neutralizing antibodies titers persist long term during the recovery phase, despite the early waning phase. The positivity rate of relatively long-term neutralizing antibodies during the convalescent phase has not been well-documented in large patient cohorts. Our findings provide evidence that 98% of survivors had sustained neutralizing antibody titers 6 months after infection.

Consistent with previous studies (Seow et al., 2020), participants who underwent mechanical ventilation or ECMO were more likely to have higher NT₅₀ than those who did not. We also found that higher BMI and fever were significantly associated with higher NT₅₀. Participants with more severe disease might have had higher viral loads, possibly inducing higher levels of viral antigen and more potent immune responses (Seow et al., 2020; Wu F. et al., 2020). Another possibility is that higher titers of antibodies could have induced more severe disease. Surprisingly, in our study, most participants who did not require oxygen support (97%) had neutralizing antibodies, suggesting that milder infections could also induce sustained immune responses. These findings might be relevant for vaccine development.

Spearman’s correlation between antibody titers revealed moderate correlations for SP IgG vs NT₅₀ (r = 0.63) and SP-total Ig vs NT₅₀ (r = 0.57). The imperfect correlation denotes that the titers of SP-IgG or SP-total Ig identify only the amount of antibodies, whereas NT₅₀ evaluates the neutralizing activity of antibodies. Hence the presence of SP IgG does not always denote the presence of neutralizing antibodies. The strengths of the correlations in our study were weaker than those of the correlations reported in the Mount Sinai study (Wajnberg et al., 2020). A stronger correlation for titers of IgG against SP by ELISA (i.e., the Mount Sinai ELISA) vs NT₅₀ (r = 0.79) at day 148 after infection was reported. Our assays for SP IgG and SP-total Ig were designed to detect antibodies against RBD antigen (Kubo et al., 2021); however, the Mount Sinai ELISA consisted of two direct serial assays against RBD and SP (Amanat et al., 2020). The differences in the assays may explain the discrepancy between the correlations. Our assays showed high sensitivity in the acute phase of the infection. The present study showed that the sensitivity (i.e., positive rates) for all the four assays was ≥84% at 6 months after the infection. In particular, SP-total Ig and NP-total Ig showed high sensitivity of 98 and 97%, respectively. Because highly sensitive and specific assays are needed to accurately estimate the frequency of previous SARS-CoV-2 infections, SP-total Ig and NP-total Ig assays are useful for seroprevalence studies.

Some limitations merit consideration. First, we could not investigate the longitudinal changes in antibody titers because of the cross-sectional design of this study. Thus, the antibody titers in our participants might decline over time. Our observation of the participants’ immune responses at a certain period strongly suggests that most exhibited potent immune responses. Second, participants recruited from the community via the media might have resulted in a healthy volunteer bias. However, the frequency (64.9%, 178/274) of participants who did not receive oxygen among hospitalized participants in our study was similar to that (61.6%, 1623/2636) reported in the COVID-19 registry in Japan (Matsunaga et al., 2020). Thus, we believe that the selection bias was not a serious concern. Third, clinical information was collected through history taking by physicians during visits to the clinics using case report forms. Because the data were not directly collected from hospitals where the participants were diagnosed or treated for COVID-19, recall bias could not be ruled out. However, because participants and clinical staffs were unaware of their antibody titers at the time of study entry, the possible bias was probably non-differential and the results of the regression analyses reported in Table 3 might have been biased toward the null. Finally, because of the inherent nature of cross-sectional study designs, we could not establish a temporal relationship between participant characteristics (e.g., oxygen support and use of steroids, etc.) and antibody titers.

In conclusion, this cross-sectional study of 376 participants revealed that most COVID-19 survivors had sustained neutralizing antibody titers 6 months after infection. Furthermore, almost all participants tested positive for SP-total Ig and NP-total Ig. These findings add to the evidence that COVID-19 survivors have humoral immune responses 6 months after infection. These data would help clinicians and public health professionals to guide their communities regarding the management and prevention of COVID-19.

The datasets presented in this article are not readily available because it is difficult to ensure de-identification of data. However, they can be available from the corresponding author on reasonable request. Requests to access the datasets should be directed to TY, yamanaka@yokohama-cu.ac.jp.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Yokohama City University. The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AG contributed to the study design, data collection, statistical analysis and interpretation of data, and the drafting and editing of the manuscript. HG, TaM, and KS contributed to the study design and data collection. KM, NO, YY, and SK performed the laboratory tests. AR and TY contributed to the study design, data collection, and supervised the analysis and preparation of the manuscript. All authors made critical revisions to the manuscript for important intellectual content and approved the final manuscript.

NO is an employee of Tosoh Corporation. YY is an employee of Kanto Chemical Co., Inc. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.