A prospective non-randomized study in parallel groups included the main group of 81 T1D patients vaccinated against influenza, and the control group of 65 children with T1D whose parents declined vaccination against influenza during the study seasons. The distribution of patients by vaccination season is presented in Table 1 .

The vaccination status of participants was confirmed by reviewing of electronic immunization program “Vaccine prophylaxis” and the child`s individual development card records. The diabetes was diagnosed according to the WHO diagnostic criteria. The WHO diagnostic criteria for diabetes include: a fasting plasma glucose level of 7.0 mmol/L (126 mg/dL) or higher, or a 2-hour plasma glucose level of 11.1 mmol/L (200 mg/dL) or higher during an oral glucose tolerance test, or a glycated hemoglobin (HbA1c) level of 6.5% or higher (20).

The study was approved by the local ethical council of the I.I. Mechnikov Research Institute of Vaccines and Serums, protocol No. 2/2014 dated 10.03.2014.

Children were enrolled in the study from the endocrinology department and the consultative and diagnostic center of the State Budgetary Healthcare Institution “Nizhny Novgorod Regional Children’s Clinical Hospital” (NRCCH, Russia) during outpatient consultations.

The characteristics of the sample in general and separately of the study groups are presented in Table 2 . The age of the patients was 12.0 (9.0-14.0) years. The gender distribution was approximately equal, with 53% girls and 47% boys. The study groups were generally comparable in terms of gender, age, duration of the T1D and most of the analyzed characteristics. However, notable differences were identified: in the group of vaccinated patients, the level of glycated hemoglobin was higher (8.9 (7.6; 10.5) % versus 7.9 (7.2; 9.0) %, p = 0.006). In the control group, no children had been vaccinated against influenza in previous seasons, whereas 56 (69%) of the vaccinated group had received influenza vaccination in the preceding season (95% CI: 58–78%).

Among the concomitant diseases, the most common were minor cardiac anomalies (in 66% of patients), biliary tract dysfunction (in 50% of patients), dysmetabolic nephropathy (in 15% of all patients). The prevalence of concomitant diseases did not differ significantly between the study groups.

According to the National Immunization Schedule of the Russian Federation, children with T1D were vaccinated with the trivalent immunoadjuvanted subunit influenza vaccine Grippol Plus (NPO Petrovax Pharm, LLC, Russia). Each dose of the vaccine contains 5 μg of hemagglutinin of each of the current epidemic strains of the influenza virus (A/H1N1, A/H3N2 and B) cultivated on chicken embryos (Abbott Biologicals B.V., Netherlands), and 500 μg of water-soluble immunoadjuvant polyoxidonium (NPO Petrovax Pharm LLC, Russia). Polyoxidonium (azoximer bromide) is registered in Russia as an immunotropic drug and has been widely used in children and adults for over 20 years (21, 22).

Epidemic strains recommended by WHO for 2014–2015 northern hemisphere vaccines were influenza type A antigen allantoic subtype (H1N1) A/California/7/2009pdm09-like (NYMC X-181), influenza type A antigen allantoic subtype (H3N2) A/Texas/50/2012-like (NYMC X-223A), influenza type B antigen allantoic B/Massachusetts/2/2012-like (NYMC BX-51B). Recommended reference viruses for 2015/2016 vaccines were allantoic influenza type A antigen A/Bolivia/559/2013 (H1N1)-like to strain A/California/07/2009 (H1N1) pdm09, allantoic influenza type A antigen subtype (H3N2) A/Switzerland/9713293/2013; allantoic influenza type B virus antigen B/Phuket/3073/2013 (23). The strain composition of the vaccines used in the study was in line with the WHO recommendations for the Northern Hemisphere and the EU decision on influenza vaccine composition for the 2014–2015 and 2015–2016 seasons. The vaccine is used according to the National Immunization Schedule in children from 6 months of age, for all age categories in adults, pregnant women, and individuals with chronic diseases (24–29).

Determination of Ab levels to strains of influenza viruses was carried out according to the hemagglutination inhibition (HI) method using influenza diagnostics of the following strains for the season 2014/2015: A/H1N1/California/07/09-like, A/H3N2/Texas/50/12-like, B/Massachusetts/2/12- like; for the season 2015/2016: study of Ab to strains A/H1N1Bolivia/559/2013-like, A/H3N2/Switzerland/9713293/2013; B/Phuket/3073/2013. (LLC “Enterprise for the production of diagnostic drugs”, Russia).

To remove non-specific hemagglutination inhibitors, the test sera were treated with receptor-destroying enzyme (RDE; Denka Seiken, Tokyo, Japan). 50 µl of the test serum was mixed with 150 µl of RDE and incubated at 37 °C overnight (19 ± 1 h), followed by 30-min inactivation at 56 °C and dilution to 1:10 with phosphate buffer. The HI reaction was performed with 0.5% chicken erythrocytes and 4 units of hemagglutination antigens. Antigens for the HI reaction were provided by the Smorodintsev Research Institute of Influenza (WHO National Influenza Center, St. Petersburg).

At the time of enrollment, children with T1D underwent baseline biochemical analyses, which included the following measurements: the level of glycosylated hemoglobin (HbA1c) was measured using an automatic DS5 Glycomat analyzer (BioChemMac, Russia) with low pressure cation exchange chromatography; blood glucose was determined using individual glucometers, and acetone in urine was measured using Dorui test strips for professional use (Dorui Industrial Company, Ltd., China).

The analysis of patient samples was carried out in the laboratory of vaccine prevention and immunotherapy of allergic diseases at the institute using certified equipment of the Center for Collective Use of the Federal State Budgetary Scientific Institution I.I. Mechnikov Research Institute of Vaccines and Serums. To remove non-specific hemagglutination inhibitors, the test sera were treated with receptor-destroying enzyme (RDE; Denka Seiken, Tokyo, Japan). 50 µl of the test serum was mixed with 150 µl of RDE and incubated at 37 °C overnight (19 ± 1 h), followed by 30-min inactivation at 56 °C and dilution to 1:10 with phosphate buffer. The HI reaction was performed with 0.5% chicken erythrocytes and 4 units of hemagglutination antigens. Antigens for the HI reaction were provided by the Smorodintsev Research Institute of Influenza (WHO National Influenza Center, St. Petersburg).

Blood samples were collected to assess Ab levels to influenza strains in both groups at three time points: at the time of enrollment, one month after vaccination, twelve months after vaccination. A titer of 1:40 or higher was considered seroprotective, while patients with titers below 1:40 were classified as seronegative The immunogenicity of the influenza vaccine was evaluated according to the efficacy criteria for adult patients aged 18–60 years established by the Committee for Proprietary Medicinal Products (CPMP/BWP/214/96), The criteria included: 1) seroprotection level: the proportion of vaccinated individuals in which the titer of hemagglutinin-inhibiting antibodies exceeded 1:40 by the 21st day after vaccination (target: >70%); 2) seroconversion level or immunological activity of the vaccine: the ratio of the number of vaccinated individuals in which the titer of hemagglutinin-inhibiting antibodies increased more than 4-fold compared to the initial level, and in which the level was not lower than 1:40 on the 21st day (target > 40%); 3) seroconversion factor (geometric mean fold rise, GMFR): an increase in the geometric mean level of hemagglutinin-inhibiting Ab on the 21st day compared to baseline, expressed as a fold increase (target: >2.5).

A vaccine is considered immunogenic if it meets at least one of these three criteria for each strain. Therefore, the determination of the seroprotection level, as well as the seroconversion level or seroconversion factor, allows us to evaluate the immunological effectiveness of the vaccine. If the vaccine meets the above immunogenicity criteria, the best epidemiological effect can be expected, especially if the seasonal circulating influenza viruses match the strain composition of the vaccine.

The study of patient samples was conducted in the laboratory of vaccination prevention and immunotherapy of allergic diseases of the institute using certified equipment of the shared use center of the Federal State Budgetary Scientific Institution “I.I. Mechnikov Research Institute of Vaccines and Serums”.

The sample size calculation was based on immunogenicity criteria, particularly a seroprotection level of at least 70%. In a pilot sample of 30 patients, the proportion of patients with a protective level of Ab (≥40) was assessed; it was 27% to the A/H1N1 strain, 34% to the A/H3N2 strain and 19% to the B strain. With an expected post-vaccination increase to at least 70%, the calculated effect sizes were 0.88 for the A/H1N1 strain, 0.74 for the A/H3N2 strain and 1.08 for the B strain. Using the pwr package and the pwr.2p.test function based on the specified information for a study power of 80%, the minimum size of each study group was calculated, which was 20 for the A/H1N1 strain, 29 for the A/H3N2 strain and 14 for the B strain (30). The total number of study groups ( Table 1 ) was 81 people in the vaccinated group and 65 patients in the control group, of which 79 patients were recruited in the 2014–2015 season. (N = 46 in the vaccinated group and N = 33 in the control group) and 67 in the 2015–2016 season. (N = 35 in the vaccinated group and N = 32 in the control group).

The Shapiro-Wilk test was used to evaluate the normality of distribution of characteristics. Significant deviations from normality were identified. Descriptive statistics of the Ab level are represented by the geometric mean titer (GMT) and its 95% confidence interval: GMT [95% CI]. Descriptive statistics of the multiplicity increase (seroconversion factor) is also represented as the geometric mean and its 95% confidence interval: GMFR [95% CI]. Descriptive statistics of other quantitative characteristics are presented by the median and interquartile range Med (Q1-Q3). The change in quantitative indicators (differences between baseline and specific time points) is represented by the median of the corresponding series of differences and its 95% confidence interval: Δ [95% CI].

To compare two unrelated samples by quantitative indicator, the Mann-Whitney test was employed. Longitudinal analysis of characteristic changes and inter-group comparisons were performed by constructing a robust linear mixed effects model (RLMEM) (31). The statistical significance of the model coefficients was determined using the Satterthwaite approximation for degrees of freedom (32). Post hoc comparisons (between groups at different time points and within groups over time) were performed by constructing the relevant contrasts using the calculated model and the emmeans package, with the Benjamini-Hochberg correction applied (33). The construction of the model for the Ab level was carried out on the data after the logarithmic transformation, for the remaining characteristics, modeling was carried out on the initial data. As fixed factors in the analysis of the formation of post-vaccination immunity, the following were determined: time after vaccination (1 month, 12 months), vaccination season (2014-2015/2015-2016), gender (male/female), age group (≥12 years/<12 years), duration of T1D (≥5 years/<5 years), prior influenza vaccination (yes/no), the initial level of Ab to the analyzed strain (seropositive/seronegative). Individual patients were assigned as random factors. Marginal R2 and conditional R2 were calculated to assess the model quality (34).

For qualitative variables, absolute and relative frequencies (in percentages) were calculated. For relative indicators, 95% CIs were determined using the Wilson method. Two groups were compared by qualitative nominal parameters using cross-tabulation analysis with the χ² test or Fisher’s exact test when expected cell frequencies were ≤5%. For paired qualitative data (e.g., pre- and post-vaccination), the McNemar test was used. To assess the strength of the relationship between vaccination and the level of seroprotection, the odds ratio was calculated, showing how many times higher the chance of achieving a seroprotective level in the vaccinated compared with the non-vaccinated. The Mantel-Henszel method was used to generalize the odds ratio (by seasons), and the Cochran-Mantel-Henszel test was applied; the homogeneity of the odds ratio was checked by the Breslow-Day test.

Statistical significance was set at p ≤ 0.05. For all multiple comparisons, the Benjamini-Hochberg correction was applied. Analyses and visualizations were performed using GraphPad Prism (v.9.3.0 License GPS-1963924) and the R statistical environment (v.3.6, License GNU GPL2).

The analysis of post-vaccination immunity (changes of the Ab level over time) was carried out taking into account gender, age of the patient, duration of T1D, and previous vaccination history. A robust linear mixed-effects model was used to analyze Ab level against the A/H1N1 strain. It was revealed that the increase in the Ab level, adjusted for interfering factors relative to the initial level, differed significantly (p <0.001) based on vaccination status ( Table 3 , all cases). It should be mentioned that the changes of the level of antibodies over time to the A/H1N1 strain a year after vaccination in the group of vaccinated patients depended on the season (p = 0.002), and in general, the level of antibodies depended on its baseline value (p <0.001).

Initially comparable in the study groups, the level of Ab to the A/H1N1 strain one month after vaccination statistically significantly increased in the vaccinated group (p <0.001 in each of the seasons and in total for two seasons), while antibody levels in the control group remained unchanged. As a result, one month after the study began, GMT of Ab in the vaccinated group was significantly higher than in the control group: 133.5 [83.8-212.9] vs. 33.1 [21.2-51.6] in the 2014–2015 season (p <0.001), 109.8 [66.3-181.9] vs. 21.3 [15.3-29.8] (p = 0.004) in the 2015–2016 season and 122.7 [87.7-171.8] vs. 26.7 [20.2-35.2] total for both seasons (p <0.001) ( Table 4 ). Changes of Аb level to the A/H1N1 ( Table 5 ) over time depending on the vaccine status for each analyzed season, in initially seronegative patients are discussed in the next section. The increase in GMT of Ab to the A/H1N1 strain one month post-vaccination relative to baseline (GMFR or seroconversion factor) was 6.2 [3.9-9.7] in the 2014–2015 season, 3.3 [2.2-4.8] in the 2015–2016 season and 4.7 [3.5-6.4] total for both seasons ( Table 6 , all cases). Twelve months after vaccination, Ab to the A/H1N1 strain in the vaccinated group declined from the one-month post-vaccination levels but remained significantly higher than baseline (p <0.001 for the 2014–2015 season, p = 0.009 for the 2015–2016 season and p <0.001 both seasons combined) and was higher than in the control group: 64.8 [43.6-96.2] vs. 26.8 [17.6-40.9] (p = 0.001) in the 2014–2015 season, 53.8 [33.9-85.5] vs. 26.5 [17.6-40.0] (p = 0.04) in the 2015–2016 season and 59.8 [44.5-80.3] vs. 26.7 [20.0-35.5] (p = 0.04) both seasons combined ( Figure 1A ). Interestingly, individuals vaccinated in the 2014–2015 season had significantly higher antibody levels one year post-vaccination compared to those vaccinated in 2015–2016 (p = 0.01). The GMT fold rise relative to baseline was in the 2014–2015 season - 3.0 [2.0-4.4] (p <0.001 vs. control), in the 2015–2016 season - 1.6 [1.1-2.3] (p = 0.05 between seasons, p = 0.37 vs. control), total for both seasons the GMT was 2.3 [1.7-3.0] (p <0.001 vs. control).

A robust mixed-effects model for changes of Ab to the A/H1N1 strain in initially seronegative participants is presented in Table 3 (SN- only).The analysis of the constructed model revealed that the level of Ab to the A/H1N1 strain in initially seronegative patients, taking into account the correction for interfering factors, were significantly dependent on vaccination status (p < 0.001 at both one month and one year post-vaccination). Additionally, the changes of Ab over time in the group of vaccinated children both one month and one year after the vaccination depended on the season (p = 0.007 and p = 0.006, respectively). In initially seronegative patients, one-month post-vaccination, the geometric mean fold rise (GMFR or seroconversion factor) was: 9.5 [5.4-16.8] times in the 2014–2015 season and 4.4 [2.3 -8.4] times in the 2015–2016 season (p = 0.05 between seasons) - Table 6 (SN- only). In both seasons, as well as across the entire study period, the increase in Ab one month after vaccination relative to baseline was statistically significant (p <0.001 in each case), the values of the indicators in the vaccinated group one month after the start of the study were significantly higher than in the control group: 97.2 [52.3-180.6] vs. 11.4 [7.6-17.2] in the 2014–2015 season (p <0.001), 46.0 [21.3-99.3] vs. 16.4 [10.9-24.8] in the 2015–2016 season (p = 0.004) and 76.5 [47.2-124.0] vs. 14.0 [10.5-18.7] in total for two seasons (p = 0.003) - Table 5 . Twelve months post-vaccination, the antibodies against the A/H1N1 strain in the initially seronegative patients declined from the one-month post-vaccination peak but remained significantly above baseline (p <0.001 in the 2014–2015 season, p = 0.02 in the 2015–2016 season and p <0.001 for both seasons combined) - Figure 1A ). Notably, in the control group there was the increase during the 2015–2016 season in GMT of Ab 12 months after the start of the study (from 10.0 [8.0-12.5] to 20.7 [11.0-39.0], p = 0.05).

Thus, twelve months after the study began, the GMT of Ab in the vaccinated group remained significantly higher than in the control group: in the 2014–2015 season (47.6 [29.1-77.8] vs. 11.4 [7.4-17.5], p = 0.02), comparable to the reference values in the 2015–2016 season (21.9 [11.3-42.6] and 20.7 [11.0 -39.0], respectively, p = 0.58) and total for both seasons (37.2 [25.0-55.2] vs. 16.0 [10.7-23.8], p = 0.04).

Among initially seronegative patients, the seroconversion factor (GMFR, or the fold increase in Ab) 12 months post-vaccination was 4.7 [2.9-7.4] in the 2014–2015 season, 2.1 [1.2- 3.6] in the 2015–2016 season (p = 0.05 between seasons) and 3.6 [2.5-5.2] total for both seasons.

Analysis of the post-vaccination immune response to the A/H3N2 strain, adjusted for confounding factors, showed that the level of antibodies to A/H3N2 after vaccination differed statistically significantly one month after vaccination (p <0.001) and one year later (p = 0.003). The dependence of the Ab level on the vaccination season (p = 0.002) and on the initial level of Ab (p <0.001) was also revealed ( Table 3 , all cases). In the 2014–2015 season, initial antibody levels were significantly higher compared to the 2015–2016 season in both study groups, and this difference remained across all subsequent time points ( Table 4 ).

One month after vaccination, there was a significant increase in GMT of Ab to the A/H3N2 strain compared to baseline both in overall (p < 0.001) and within each season (p <0.001 - season 2014 - 2015, p <0.001 - season 2015 - 2016), while in the control group one month after the start of the study there was a slight decrease in the Ab level to the A/H3N2 strain compared to baseline ( Figure 1B ). The seroconversion factor (GMFR)—the fold increase in GMT one-month post-vaccination—was: 3.6 [2.8-4.6] in the 2014–2015 season, 2.9 [1.9-4, 4] in the 2015–2016 season and 3.2 [2.5-4.1] overall in both seasons - Table 6 (all cases). As a result of the described changes, one month after vaccination/start of observation, the GMT of Ab antibodies to the A/H3N2 strain in the group of vaccinated persons were significantly higher than in the non-vaccinated group: 152.9 [108.0-216.6] versus 41.7 [27.9-62.4] in the 2014–2015 season (p <0.001), 46.0 [37.7-56.0] versus 9.8 [8.1-11.8] in the 2015–2016 season (p <0.001) and 91.0 [70.9-116.7] versus 20.4 [15.4-27.1] overall (both seasons) (p <0.001).

Twelve months after vaccination, GMT of Ab to the A/H3N2 strain in the vaccinated group showed a statistically significant decline compared to the one-month level (p = 0.02 - season 2014-2015, p <0.001 - season 2015-2016, p <0.001 - overall (both seasons). However, relative to baseline, antibody levels remained elevated in the 2014–2015 season (p = 0.001) but were comparable to the baseline in the 2015–2016 season (p = 0.67). Compared to the control group, the Ab level to A/H3N2 strain 12 months after vaccination was statistically significantly higher in the 2014–2015 season (95.9 [66.9-137.3] vs. 53.7 [33.1-87.1], p = 0.005) and comparable in the 2015–2016 season (15.8 [11.4-21.7] and 14.1 [9.7-20.6], respectively, p = 0.19). Overall (both seasons) 12 months after vaccination, the level of antibodies to the A/H3N2 strain was 44.0 [32.1-60.1], which is higher than the initial level (28.7 [21.2-38.7], p = 0.01) and higher than in the control group (27.8 [19.8-39.2], p = 0.01).

Twelve months post-vaccination, the seroconversion factor (GMFR) relative to baseline was 1.8 [1.1-3.0] in the 2014–2015 season (p = 0.02 compared to the control group), 1.2 [0.8-1.8] in the 2015–2016 season (no significant difference from the control group, p = 0.19), and overall (both seasons): 1.5 [1.1–2.1] vs. 0.8 [0.6–1.1] in the control group (p = 0.02).

The model describing the formation of post-vaccination immunity (changes of the Ab level over time) to the A/H3N2 strain in the initially seronegative participants is presented in Table 3 (only SN-). It was revealed that the Ab level to the A/H3N2 strain in initially seronegative patients depended mainly on vaccination (p <0.001 - one month, p = 0.002 - twelve months post-vaccination). It’s also worth mentioning that the changes of Ab over time in the group of vaccinated children one year after the vaccination depended on the season (p = 0.008). Initially comparable GMT of Ab to the A/H3N2 strain between the study groups and seasons one month after vaccination significantly increases in the vaccinated group (p <0.001 - in the 2014–2015 season, p <0.001 – in the 2015–2016 season and p <0.001 total for two seasons) - Table 5 . It is worth noting that this increase (seroconversion factor) was more intense in the 2014–2015 season compared to the 2015–2016 season (10.1 [4.2-24.0] vs. 4.2 [3.4-5, 2], p = 0.04). The monthly seroconversion factor for the two seasons was 5.6 [4.0-7.7] - Table 6 (only SN-).

Thus, 1 month after vaccination/study initiation, the GMT of antibodies to the A/H3N2 strain in participants without an initial protective level was significantly higher in the group of vaccinated participants, compared to the control group: 91.9 [37.5-225.3] vs. 8.9 [2.8-28.6] in the 2014–2015 season (p <0.001), 45.7 [36.6-57.1] vs. 9.4 [7.6-11.6] in the 2015–2016 season (p <0.001) and 57.4 [41.7-79.2] vs. 9.3 [7.3-11.9] across both seasons combined (p <0.001) - Figure 1B ).

Changes of the Ab levels over time to the A/H3N2 strain in initially seronegative patients 12 months after vaccination depended on the study season. In the 2014–2015 season, antibody levels in the vaccinated group remained high (with no significant changes compared to the one-month level, p = 0.08), exceeding baseline values (p < 0.001) and remaining higher than the GMT of Ab to the A/H3N2 strain in the unvaccinated group (p < 0.001). In contrast, in the 2015–2016 season, the GMT of Ab against the A/H3N2 strain declined significantly 12 months after vaccination compared to the one-month level (p < 0.001), returning to baseline values (p = 0.38) and becoming comparable to levels in the unvaccinated group (p = 0.77). Overall, across both seasons, 12 months after vaccination, the level of Ab to the A/H3N2 strain in the initially seronegative study participants was 24.3 [16.6-35.8], which was higher than the initial level (10.3 [8.9-12.0], p = 0.02) and higher than in the control group (12 months after the start of the study - 13.8 [8.8-21.7], p = 0.04).

The seroconversion factor to the A/H3N2 strain in initially seronegative patients 12 months after vaccination, relative to the initial level, was 6.4 [2.6-15.3] in the 2014–2015 season and 1.5 [1.0-2.2] in the 2015–2016 season, with statistically significant differences between seasons (p = 0.005). The overall seroconversion factor (GMFR) across both seasons 12 months after vaccination, relative to the initial level, was 2.3 [1.5–3.5], with no statistically significant differences compared to the unvaccinated group (p = 0.20).

The model describing changes in Ab levels over time against strain B is presented in Table 3 (all cases). Analysis of the constructed model showed that the post-vaccination immune response depended on the patient’s age, with a higher response observed in children over 12 years old (29.2 [24.5-34.8] vs. 18.1 [15.7-20.9], p <0.001), the initial Ab level (p <0.001), and whether the individual was vaccinated in the current season (p <0.001 after 1 month, p = 0.001 after 12 months).

One month after vaccination, the GMT of Ab against the B strain increased significantly in the vaccinated group compared to baseline (p <0.001 in each season and total across both seasons) - Table 4 . Twelve months after vaccination, there was a decline in the Ab levels to B strain compared to the one-month level (p <0.001 for the 2014–2015 season, p = 0.003 for the 2015–2016 season and p <0.001 for both seasons combined). However, GMT values remained above baseline (p = 0.003 for the 2014–2015 season, p = 0.001 for the 2015–2016 season, p <0.001 for both seasons combined). In the control group, no statistically significant changes in GMT of Ab to B strain were observed throughout the study period. As a result, one month after vaccination/start of the study, the GMT of Ab to strain B was significantly higher in the vaccinated group compared to the control group (53.3 [37.8-75.1] vs. 9.6 [6.9-13.3] (p <0.001) in the 2014–2015 season, 91.9 [65.7-128.6] vs. 14.8 [10.6-20.7] (p <0.001) in the 2015–2016 season and 66.7 [52.1-85.4] vs. 11.9 [9.4-15.0] (p <0.001) both seasons combined). Twelve months after vaccination, GMT antibody levels in the vaccinated group remained significantly higher than in the control group (27.4 [20.1-37.5] versus 7.9 [6.0-10.5] (p <0.001) in the 2014–2015 season, 44.2 [34.3-56.9] vs. 15.8 [12.3-20.3] (p <0.001) in the 2015–2016 season and 33.6 [27.2-41.6] vs. 11.1 [9.1-13.6] (p <0.001) in total for two seasons) - Figure 1C ).

The seroconversion factor (GMFR) in the vaccinated group one month after vaccination compared to the baseline level was 3.2 [2.2-4.7] in the 2014–2015 season (p <0.001 vs. control), 4.1 [2.5-6.6] in the 2015–2016 season (p <0.001) and 3.6 [2.7-4.8] for both seasons combined (p <0.001); twelve months after vaccination - 1.7 [1.2-2.2] in the 2014–2015 season (p = 0.008 vs. control), 2.0 [1.4-2.8] in the 2015–2016 season (p = 0.01) and 1.8 [1.4-2.2] for both seasons combined (p <0.001) – see Table 6 (all cases).

The analysis of the changes in Ab levels over time to strain B in the initially seronegative study participants is given in Table 3 (only SN-). It was revealed that the level of antibodies to strain B in initially seronegative patients depended only on the fact of vaccination (p <0.001 after 1 month, p = 0.001 after 12 months). The changes of the Ab level over time to strain B in initially seronegative participants is characterized by a statistically significant increase one month after vaccination compared to baseline (p <0.001 for both seasons and in total for two seasons) with a subsequent decrease after 12 months (p <0.001 for both seasons and in total across two seasons), while the GMT of Ab to B strain and one year after vaccination remained above the baseline (p <0.001 for both seasons and in total for two seasons) - Table 5 . When compared to the control group, GMT of Ab levels to B strain in initially seronegative patients in the vaccinated group were significantly higher both one month (40.9 [26.2-64.1] vs. 7.3 [5.6- 9.6] in the 2014–2015 season (p <0.001), 73.7 [40.4-134.7] vs. 10.4 [7.4-14.6] in the 2015–2016 season (p <0.001) and 50.7 [35.5-72.3] vs. 8.5 [6.9-10.5] total for two seasons (p <0.001)) and one year post-vaccination (19.1 [13.0 -28.0] vs. 6.0 [5.2-6.9] in the 2014–2015 season (p <0.001), 31.3 [23.0-42.6] vs. 14.6 [10.8- 19.9] in the 2015–2016 season (p <0.001) and 22.8 [17.4-29.9] vs. 8.8 [7.2-10.7] total for two seasons (p <0.001)) - Figure 1C ).

Among initially seronegative patients, the seroconversion factor (GMFR, or the fold increase in Ab) one month after vaccination was 4.7 [2.9-7.5] in the 2014–2015 season (p <0.001 compared to the control), 8.7 [4.1-18.3] in the 2015–2016 season (p <0.001), and 5.9 [4.0-8.7] total for two seasons (p <0.001); 12 months post-vaccination was 2.2 [1.5-3.2] in the 2014–2015 season (p <0.001), 3.7 [2.3-5.8] in the 2015–2016 season (p = 0.02), and 2.6 [2.0-3.5] total for two seasons (p <0.001) - Tables 6 and 7 (only SN-).

An analysis of the vaccine’s compliance with the CPMP/BWP/214/96 criteria showed that the level of seroprotection one month after vaccination in total for the two seasons 2014–2015 and 2015–2016 was 80 [70–87]% to A/H1N1, 89 [80–94]% to A/H3N2 and 77 [66-84]% to strain B, which met the criteria (seroprotection level target: >70%) and was significantly higher than the values in the control group (46 [35-58]%, 37 [26-49]% and 18 [11-30]%, respectively, p <0.001 for each strain). A similar pattern (compliance with the criterion and statistically significant excess of the indicator values in the control group) is typical for each season considered - Table 7 (all cases).

The seroconversion rate also met the CPMP criteria (target: > 40%) and significantly exceeded the values in the control group: 57% [46-67%] vs. 9% [4-19%] for A/H1N1 (p <0.001), 49% [39-60%] vs. 3% [1-11%] for A/H3N2 (p <0.001), and 46% [35-56%] vs. 9% [4-19%] for strain B (p <0.001).

The seroconversion factor for two seasons to the A/H1N1 strain in the vaccinated group was 4.7 [3.5-6.4] vs. 1.1 [0.9-1.3] in the control group (p <0.001), to the A/H3N2 strain - 3.2 [2.5-4.1] vs. 0.6 [0.5-0.8] (p <0.001), for strain B - 3.6 [2.7-4.8] vs. 0.9 [0.7-1.1] (p <0.001), meeting CPMP criteria (which must be at least 2.5-fold) in total and in each of the seasons alone- Table 6 (all cases).

Also, the level of seroprotection, factor and level of seroconversion were calculated separately for seronegative patients. One month after the vaccination, the level of seroprotection (combined across both seasons) to strains A/H1N1, A/H3N2 and B was 68 [54-80]%, 80 [67-89]% and 66 [52-78]%, respectively, seroconversion rate -70 [56-81]%, 76 [62-86]% and 62 [47-74]%, seroconversion factor-7.4 [4.8-11.5], 5.6 [4, 0-7.7] and 5.9 [4.0-8.7]. Detailed calculations for each season, as well as efficacy indicators 12 months post-vaccination, are presented in Table 6 and Table 7 .

It should be noted that there is a fairly strong variability in the performance indicators depending on the season, including the results of comparisons with the control group. In order to eliminate this variability and to generalise the results obtained, odds ratios were calculated, which indicate how many times the chance of achieving a seroprotective level of Ab (above 40) is greater in the vaccinated group than in the unvaccinated group. To aggregate the odds ratios across different epidemiological seasons, the Mantel-Henszel odds ratio was calculated, with the method’s conditions verified using the Breslow-Day test. The relative odds of achieving seroprotection one month after vaccination/study entry and of maintaining a protective Ab level 12 months in the vaccinated group (compared to unvaccinated group) are shown in Figure 2 .

Regardless of the epidemiological season, vaccination significantly increased the chance of reaching seroprotective Ab level one month after vaccination: the combined odds ratios (Mantel-Henszel method) were: 4.7 [2.9-9.7] for the A/H1N1 strain (χ²MH = 16.4, p < 0.001), 15.8 [5.9-41.4] for the A/H3N2 strain (χ²MH = 44.0, p < 0.001) and 14.8 [6.5-33.6] for strain B (χ²MH = 46.2, p < 0.001).

Another situation was observed when analyzing the chances of maintaining the seroprotective level 12 months after the vaccination. The highest persistence was shown by Ab to B strain, with the odds of maintaining seroprotection 7.2 times higher than in the non-vaccinated group (OR = 7.2 [3.2–16]), independent of the season. Ab resistance to the A/H1N1 strain was season-dependent (it was lower in the 2015–2016 season), the combined Mantel-Henszel odds ratio across seasons was 2.5 [1.3–5] (χ²MH = 6.5, p = 0.01). Ab to the A/H3N2 strain demonstrated the least persistence: one year post-vaccination, the odds of maintaining seroprotection were equivalent to the non-vaccinated group (OR = 1.0 [0.5–2.3], χ²MH = 0.02, p = 0.89).