2.1 Study design and patient enrollment

We enrolled a section cohort of patients with cancer, mainly patients with breast and liver cancers who received a vaccine before having an omicron subvariant infection during the latest BF.7 and BA.5.2 waves in China in December 2022. The patients were admitted to the Fifth Medical Center of the Chinese PLA General Hospital for cancer treatment between February 15, 2023 and March 15, 2023. Finally, 58 and 81 unvaccinated and vaccinated patients with breast cancer, respectively, and 47 unvaccinated and 62 unvaccinated and vaccinated patients with liver cancer, respectively, were included (Figure S1). All patients with cancer were infected during the omicron surge in December 2022 and were diagnosed with mild or moderate COVID-19 disease. The diagnostic and classification criteria for coronavirus infection were based on the “Guidelines for Prevention and Control of Coronavirus Disease 2019 (Version 9).”²⁶ All patients signed informed consent. This study was approved by the Ethics Committee of the Fifth Medical Center of the Chinese PLA General Hospital (ethical approval no. ky-2021-7-9-1).

2.2 Clinical and laboratory parameters collection

We collected data on demographics, comorbidities, vaccinations, treatments, tumor node metastasis (TNM) stage, and lymphocyte counts at admission for cancer treatment at our center from the patients' medical records.²⁷ Two other physicians checked all data separately, and a third researcher resolved any differences in interpretation between the two reviewers.

2.3 Definitions of vaccination

The vaccination status was categorized into three groups: unvaccinated, fully vaccinated, and booster doses. The vaccine administered to patients in our study was mainly inactivated vaccine. Patients who received two doses of the inactivated vaccine were defined as fully vaccinated, and those who received three doses of the inactivated vaccine were defined as having received a booster dose. The interval between the last dose and breakthrough infection and the time from COVID-19 infection to hospitalization were counted separately.

2.4 Serum samples collection for anti-spike RBD IgGs and 50% inhibiting concentration assay

In the section cohort, serum samples were collected 70–80 days after the surge of Omicron infection in the winter of 2022. They were classified into three groups based on the vaccination strategy: unvaccinated, fully vaccinated, and booster doses.

The pseudotyped virus neutralization assay was similar to those of previous reports,²⁸ and the neutralization ability of serum samples against preceding (wild-type), circulating Omicron (BA.4-BA.5 and BF.7) and a new subvariant (XBB.1.5) were measured. Briefly, in this experiment, the pseudovirus neutralization detection method was used to use an HIV lentiviral vector, a pseudovirus based on the surface expression of SARS-CoV-2 S protein, which can bind to highly-expressed angiotensin-converting enzyme 2 (ACE2) of the cells and invade and integrate genes into the cells. The pseudoviral gene was modified to carry the luciferase reporter gene, which can express the luciferase protein, and can emit light color by adding the corresponding luciferase substrate to the detection system. Finally, the chemiluminescence detection function module of the microplate reader was used to detect the chemiluminescence signal value (RLU), the detection wavelength was set to full wavelength, and the reading time was set to 500 ms. The relative luminescence signal value (RLU) is inversely proportional to the amount of anti-SARS-CoV-2 S protein-neutralizing antibodies in the sample to be tested. Evaluation of pseudovirus-infected cell activity was performed by detecting luciferase activity. Enzyme-linked immunosorbent assay (ELISA) was used to quantitatively detect s-IgG antibodies against SARS-CoV-2 Spike RBD (original and Omicron BA.4-BA.5, BF.7, BA.4.6, and XBB.1.5 strains) protein in human plasma using previously-described quantitative methods.²⁹

2.5 Statistics

All statistical analyses were performed using the SPSS Statistics 24. Data were visualized using GraphPad Prism 8. Categorical variables are presented as counts and percentages, and continuous variables are presented as mean ± standard deviation. Fisher's exact test and χ² test were used to compare categorical variables. Analysis of variance was used to compare continuous variables. Statistical significance was set at p < 0.05. Multivariate analysis was performed using a liner regression model to identify the factors that may be associated with anti-spike RBD IgGs and 50% pseudovirus neutralization titer (pVNT₅₀). All potential variables were screened before the multivariate analysis. All qualified variables were subjected to single-factor regression, and variables with p < 0.1 were included in the subsequent multivariable regression analysis. The variance inflation factor was calculated for each iteration to avoid multicollinearity. Due to the strong correlation between sex and tumor type, sex variables were excluded from the multivariate analysis.

3.1 Demographic characteristics of patients

In our cohort, we included 58 unvaccinated and 81 vaccinated patients with breast cancer, 47 unvaccinated and 62 vaccinated patients with liver cancer. Data on age, sex, BMI, comorbidities, cancer disease status, and antitumor treatments were collected (Table 1). In addition, we also recorded the vaccinations, disease severity of omicron infection, clinical symptoms, antiviral treatment and the date of infection (Table 1).

The TNM stage of patients was considered as an important factor that leads to the vaccine hesitancy. There was no difference in the TNM staging between breast cancer and liver cancer patients who have been vaccinated; however, the number of unvaccinated patients with breast cancer at stage IV (52/58) was higher than those patients with liver cancer (28/47). In breast cancer patients, the tumor stage is closely related to the tumor-associated therapies, while treatments do not differ in liver cancer patients.

Among the vaccinated patients, the rates of fully vaccination and the booster doses were 19.4% (12/62) and 80.6% (50/62) respectively in patients with liver cancer, and 19.8% (16/81) and 80.2% (65/81) respectively in patients with breast cancer. We also recorded symptoms after infection, medications, and the time between infection and the last vaccination. As shown that the breast cancer patients are more likely to experience muscle soreness and fatigue, and the proportions of receiving traditional Chinese medicine (Lianhua Qingwen capsules) for treatment was higher than that in liver cancer group. No patients required respiratory support or ICU admission, and the severity of COVID-19 was significantly lower in the vaccinated patients than the vaccinated patients.

The period between the last dose and breakthrough infection in patients with liver cancer was 347, IQR (273, 414), and the period between the last dose and breakthrough in patients with cancer was 372, IQR (324, 416) days. The average time from the COVID-19 infection to hospitalization for cancer treatment was 70.6 days for patients with liver cancer and 74.3 days for those with breast cancer (Table 1).

We conducted a further analysis of peripheral lymphocyte subsets, including CD4⁺ T, CD8⁺ T, CD19⁺ B, and NK cells, between the two groups. In patients with breast cancer, the count of CD19⁺ B cells in unvaccinated individuals was significantly higher than that in vaccinated patients, while CD4 and CD8 counts were significantly lower in unvaccinated breast cancer patients compared to vaccinated ones. For liver cancer, only the CD4 count in unvaccinated individuals was significantly lower than that in vaccinated patients. However, the CD4, CD8, and NK cell counts were even higher in breast cancer patients than in liver cancer patients in spite of vaccination status, highlighting the disparity of immune cell profiles between the two cancer groups. (Table 2).

3.2 Evaluation of anti-spike RBD IgGs antibody level between the unvaccinated and vaccinated patients with cancer

In this study, we compare the levels of anti-spike RBD IgGs for the preceding (wildtype), circulating Omicron (BA.4-BA.5, and BF.7), and new subvariants (XBB.1.5) after breakthrough infection. The vaccinated patients with cancer had higher titers of anti-spike IgGs against WT, BA.4-BA.5, BF.7, and XBB.1.5 than unvaccinated patients (Figure 1A), and patients who received booster doses had much higher titers than fully vaccinated patients (Figure 1B). We compared the efficacy of booster doses on the level of anti-spike RBD IgGs in patients with different types of cancer, and found that patients with liver cancer receiving a three-dose inactivated vaccine had the highest titers of anti-spike RBD IgGs against WT, BA.4-BA.5, BF.7, and XBB.1.5, within the whole population (Figure 1C). Furthermore, when patients were divided by cancer type, unvaccinated patients with breast cancer had higher levels of anti-spike RBD IgGs against BF.7 and XBB.1.5 than patients with liver cancer after Omicron subvariant infection (p = 0.017, p = 0.005). For booster-vaccinated patients with cancer, patients with liver cancer had significantly higher titers of anti-spike RBD IgGs against BA.4-BA.5, BF.7, and XBB.1.5, than patients with breast cancer after breakthrough infection (p = 0.013, p = 0.017, p = 0.003) (Figure 1C).

We also separately analyzed the effects of anti-tumor treatments on the anti-spike RBD IgGs in the two types of patients with cancer and found that vaccinated patients with breast cancer who received endocrine therapy were more prone to produce higher levels of anti-spike IgGs against BA.4-BA.5, BF.7, and XBB.1.5, and PD-1 treatment had no effect on anti-spike RBD IgG production in patients with liver cancer despite vaccination (Figure S2).

Multivariate analysis revealed that vaccine booster was the only factor associated with high titers of anti-spike RBD IgGs against WT, BA.4, BA.5, BF.7, and XBB.1.5 in all patients with cancer (p < 0.05) (Table S1–S4).

3.3 Evaluation of neutralizing antibody titers in different patients with cancer

To identify the neutralizing efficacy of serum samples against SARS-CoV-2 variants in vaccinated patients with cancer after breakthrough infection, we first analyzed the neutralizing antibodies against wild-type, BA.4-BA.5, BF.7, and XBB.1.5, in unvaccinated and vaccinated patients after breakthrough infection. The results showed that vaccinated patients with cancer had higher levels of 50% pseudovirus neutralization titer (pVNT₅₀) against wild-type, BA.4-BA.5, and BF.7 than unvaccinated patients, but the unvaccinated patients had higher titers against XBB.1.5 than vaccinated patients (Figure 2A). Second, for those patients who received a booster doses of ancestral SARSCoV-2 produced slightly higher levels of pVNT₅₀ against WT, BA.4-BA.5, BF.7 than fully vaccined ones in spite of no significance (Figure 2B). The above results supposed that the omicron variant vaccine booster may be better than the inactivated ancestral vaccine booster for providing omicron protection.¹⁴ Third, we also analyzed and compared the levels of pVNT₅₀ in breast cancer and liver cancer group received different dose of vaccination. In the breast cancer patients received fully vaccination did not produce higher level of pVNT₅₀ against BA.4-BA.5 and BF.7 compared to unvaccinated patients after breakthrough infection. However, pVNT₅₀ levels were notably increased in patients who received a booster dose compared to unvaccinated patients (Figure 2C). Within the liver cancer group, patients who received both fully and booster doses shown a significant elevation in pVNT₅₀ against BA.4-BA.5 and BF.7 compared to unvaccinated patients (Figure 2C). Overall, all patients shown slightly higher pVNT₅₀ levels after receiving a booster dose than fully vaccination. Notably, only unvaccinated breast cancer patients had a cross-neutralization against XBB.1.5 following infection, which was not observed in patients with liver cancer (Figure 2C). These results suggested that a booster dose with the ancestral vaccine can only slightly enhance neutralizing antibody levels against the circulating Omicron variant, yet it does not confer cross-protection against the new subvariant. (Figure 2C).

3.4 Association of anti-spike RBD IgGs and neutralizing antibodies in different patients with cancer

The associations between the anti-spike RBD IgGs and neutralizing antibodies (pVNT₅₀) against WT, BA.4-BA.5, BF.7, and XBB.1.5 were separately recorded among patients based on cancer type and vaccination. For unvaccinated patients, positive associations between anti-spike RBD IgGs (BA.4-BA.5, BF.7, and XBB.1.5) and the corresponding neutralizing antibodies were found only in unvaccinated patients with breast cancer (Figure 3A). Positive associations between anti-spike RBD IgGs against WT, BA.4-BA.5, and BF.7, and the corresponding neutralizing antibodies were found in all vaccinated patients after breakthrough infection (Figure 3B,D). Furthermore, a positive association between anti-spike RBD IgG against XBB.1.5 and the corresponding neutralizing antibody was observed in vaccinated patients with liver cancer (Figure 3C). Consistent with the expected results, no association was observed between anti-spike RBD IgG against the WT and the corresponding neutralizing antibodies in unvaccinated patients after omicron infection.