Immune responses and clinical outcomes after COVID-19 vaccination in patients with liver disease and liver transplant recipients

Background & Aims Comparative assessments of immunogenicity following different COVID-19 vaccines in patients with distinct liver diseases are lacking. SARS-CoV-2-specific T-cell and antibody responses were evaluated longitudinally after one to three vaccine doses, with long-term follow-up for COVID-19-related clinical outcomes. Methods A total of 849 participants (355 with cirrhosis, 74 with autoimmune hepatitis [AIH], 36 with vascular liver disease [VLD], 257 liver transplant recipients [LTRs] and 127 healthy controls [HCs]) were recruited from four countries. Standardised immune assays were performed pre and post three vaccine doses (V1-3). Results In the total cohort, there were incremental increases in antibody titres after each vaccine dose (p <0.0001). Factors associated with reduced antibody responses were age and LT, whereas heterologous vaccination, prior COVID-19 and mRNA platforms were associated with greater responses. Although antibody titres decreased between post-V2 and pre-V3 (p = 0.012), patients with AIH, VLD, and cirrhosis had equivalent antibody responses to HCs post-V3. LTRs had lower and more heterogenous antibody titres than other groups, including post-V3 where 9% had no detectable antibodies; this was heavily influenced by intensity of immunosuppression. Vaccination increased T-cell IFNγ responses in all groups except LTRs. Patients with liver disease had lower functional antibody responses against nine Omicron subvariants and reduced T-cell responses to Omicron BA.1-specific peptides compared to wild-type. 122 cases of breakthrough COVID-19 were reported of which 5/122 (4%) were severe. Of the severe cases, 4/5 (80%) occurred in LTRs and 2/5 (40%) had no serological response post-V2. Conclusion After three COVID-19 vaccines, patients with liver disease generally develop robust antibody and T-cell responses to vaccination and have mild COVID-19. However, LTRs have sustained no/low antibody titres and appear most vulnerable to severe disease. Impact and implications Standardised assessments of the immune response to different COVID-19 vaccines in patients with liver disease are lacking. We performed antibody and T-cell assays at multiple timepoints following up to three vaccine doses in a large cohort of patients with a range of liver conditions. Overall, the three most widely available vaccine platforms were immunogenic and appeared to protect against severe breakthrough COVID-19. This will provide reassurance to patients with chronic liver disease who were deemed at high risk of severe COVID-19 during the pre-vaccination era, however, liver transplant recipients had the lowest antibody titres and remained vulnerable to severe breakthrough infection. We also characterise the immune response to multiple SARS-CoV-2 variants and describe the interaction between disease type, severity, and vaccine platform. These insights may prove useful in the event of future viral infections which also require rapid vaccine development and delivery to patients with liver disease.


Introduction
The rapid development and deployment of vaccinations against SARS-CoV-2, alongside a degree of naturally acquired immunity from past infection, has transformed the landscape of the COVID-19 pandemic.At a population level, vaccination has been shown to reduce SARS-CoV-2 infection and protect against hospitalisation and death from severe COVID-19. 1,2atients and methods

Study design and sampling protocol
We evaluated the humoral and cellular immune response after COVID-19 vaccination delivered as part of clinical care to European patients with cirrhosis, AIH, or VLD, or LTRs.Patients were prospectively recruited across two multicentre consortia; the EASL supported COVID-Hep vaccine network and the UK OCTAVE (Observational Cohort trial T-cells, Antibodies and Vaccine Efficacy in SARS-CoV-2) study.Both studies had identical inclusion criteria; >16 years of age, eligibility for COVID-19 vaccination with BNT162b2, mRNA-1273, or ChA-dOx1 nCoV-19 (AZD1222, ChAdOx1) vaccine platforms, and an anticipated life expectancy >6 months.Healthy controls were recruited via the UK PITCH (Protective Immunity from T Cells in Healthcare workers) consortium.All studies included clinical and demographic data collection at baseline followed by longitudinal collection of serum and peripheral blood mononuclear cells (PBMCs) at standardised timepoints throughout the vaccination schedule, although sample availability at each timepoint was variable.5][16] However, these studies included: i) patients who had received a single vaccine type such that comparisons of immunogenicity and vaccine effectiveness between vaccine types could not be made, 16 ii) patients who received only two vaccine doses and no booster vaccines 14,15 and iii) did not include any data on SARS-CoV-2 infection rates and disease severity data. 15,16hical and regulatory approvals All centres involved in the EASL supported COVID-Hep vaccine registry recruited participants through local ethics approvals.All studies were conducted in compliance with relevant ethical regulations for work with human participants according to the principles of the Declaration of Helsinki (2008) and written informed consent was obtained for all included participants.Additional methods are available in the supplementary materials.

Outcome measures and sample size calculations
The two primary outcomes of this study were levels of anti-SARS-CoV-2 IgG antibodies and the magnitude of the T-cell responses to wild-type (WT) SARS-CoV-2 peptides following COVID-19 vaccination.The study was powered to detect a difference in SARS-CoV-2 IgG anti-S antibody titres 28 days after V2 between disease cohorts (cirrhosis, AIH, VLD, LTR) and HCs.Sample size calculations were performed a priori based on available data at the time of study conception relating to COVID-19 vaccine immunogenicity in phase III trials and historic data showing diminished vaccine responses to a range of other viruses in LTRs and patients with cirrhosis.To allow for comparisons between vaccine platforms and assuming a 20% reduction in antibody titres in patients relative to controls we estimated that at least 100 patients per disease group would be required to detect a difference with 90% power and an alpha of 0.05.Secondary outcomes included the magnitude of the T-cell responses specifically to SARS-CoV-2 Omicron BA.1 peptides, IgG binding and inhibition of angiotensin-converting enzyme 2 (ACE2) binding to SARS-CoV-2 VoC, and rates and severity of breakthrough SARS-CoV-2 infection after COVID-19 vaccination.

Clinical phenotyping and definitions
Clinical data for all participants were uploaded electronically from participating sites using REDCap (Research Electronic Data Capture) databases hosted by the University of Oxford or University of Birmingham, UK.Clinical data comprised information on demographics, vaccination type, comorbidities, and disease-specific phenotyping including Child-Pugh (CP) class and aetiology of cirrhosis, and type and dose of immunosuppression for patients with AIH and LTRs.Previous SARS-CoV-2 infection was defined as a patient-reported episode of confirmed COVID-19 or a positive antibody or T-cell response to SARS-CoV-2 nucleocapsid antigen at the first timepoint at which a patient was sampled.Nucleocapsid antibody titres were additionally assessed at post-V1, and post-V3 timepoints.Model for end-stage liver disease (MELD) score included serum sodium, creatinine, bilirubin and international normalised ratio. 17

Anti-SARS-CoV-2 Ig analysis
For all participants, the magnitude of anti-SARS-CoV-2 antibodies was measured using identical Roche Elecsys ® Anti-SARS-CoV-2-S (anti-S) and Roche Elecsys ® Anti-SARS-CoV-2-N (anti-N) assays.The Roche Anti-SARS-CoV-2-S assay measures the presence and the amount of serum antibodies to the spike receptor binding domain (RBD) antigen of SARS-CoV-2.Seroconversion was manufacturer defined as anti-S antibodies > − 0.8 U/ml and anti-N antibodies >1U/ml.The upper limit of detection for anti-S antibodies was 25,000 U/ml.Low response was defined as <380 U/ml as per. 14Assays for all samples were completed at the University of Hamburg or the UKHSA Laboratories at Porton Down.

IFNc T-cell ELISpot assay
IFNc ELISpots were performed using Human IFNc ELISpot Basic kit (Mabtech) as previously described. 18Wells included 200,000 thawed PBMCs and stimulation conditions included overlapping peptide pools (18 mers with 10 amino acid overlap), negative control (DMSO only) or positive control (CEF and concanavalin A).IFNc ELISpots for liver disease groups were all performed at one laboratory at the University of Oxford.IFNy ELISpots for healthy controls were performed separately at the University of Oxford.Samples were only included in the analysis if the IFNc response to DMSO was <50 spot-forming units (SFU)/10 6 PBMCs and positive responses were detected in the positive controls.Additional methods are available in the supplementary materials.

Breakthrough SARS-CoV-2 infection after COVID-19 vaccination
Information regarding breakthrough SARS-CoV-2 infection after vaccination was also collected by screening electronic hospital records and/or contacting individual patients and included date of infection and COVID-19 severity.Data was collected up until December 2022.Severe COVID-19 was defined according to the World Health Organisation classification or based on hospitalisation status. 19Additional methods are available in the supplementary materials.

Statistical methods
Descriptive statistics are presented as n (%) or median (IQR), unless otherwise indicated.A nominal 2-sided 5% significance level was adopted unless otherwise stated.Kruskal-Wallis test followed by post hoc Dunn's testing with Holm-Bonferroni multiplicity adjustment was used for multiple pairwise comparisons.Mann-Whitney U test, Wilcoxon matched-paired signed rank test, Pearson's Chi-Square, and Fisher's exact test were used where required as per figure legends.Multiple comparisons corrections were made as per individual figure legends (Holm-Bonferroni adjustments).Spearman's correlations or linear regressions of log 10 transformed values investigated relationships between variables of interest.Spearman's correlations were used to quantify the association between variables and linear regression models were used to model effects of clinical or immunological factors on anti-RBD Ig responses.Multivariable models (linear or logistic as per legends) were used to model multivariate effects of clinical and immunological factors on log 10 transformed anti-RBD Ig responses (linear) or odds ratio of seropositivity (>0.8 AU/ml, logistic) at post-v2 in the whole cohort or in specific disease groups.Models were created as per figure legends.Statistical analyses were performed using R (v4.
Across the entire liver disease cohort, the median age was 60 years (IQR 52-68), 425 (59%) were male, and 76 (11%) had previous evidence of SARS-CoV-2 infection.In the HC cohort, the median age was 36 years (IQR 25-45), 40 (31%) were male, and 40 (31%) had evidence of previous SARS-CoV-2 infection.Nineteen patients became newly positive for nucleocapsid antibody at the post-V3 timepoint.Clinical information for each specific disease category is presented in Table 1.For the analysis, 21 (29%) patients with AIH and concurrent cirrhosis were included in the AIH group and not the cirrhosis group.

Antibody responses across liver disease phenotypes and vaccine platforms
Longitudinal assessment of antibody responses using the Roche anti-RBD assay across all liver disease phenotypes and HCs is presented in Fig. 1A.Prior SARS-CoV-2 infection was Responses after COVID-19 vaccination associated with a significant increase in anti-RBD titres across all disease groups (Fig. S1) and therefore previously infected individuals were removed from the primary analysis (n = 76 with liver disease and n = 40 HCs excluded).Across the total liver disease cohort, there was a stepwise incremental increase in median antibody titres after each consecutive vaccine dose (6.24 U/ml [0.4-44.9]post-V1 vs. 846 U/ml [158-2,653] post-V2 vs. 12,746 U/ml [2,508-25,000] U/ml post-V3; p < 0.0001).This observation remained significant after excluding 19 patients who had become newly positive for nucleocapsid antibody between enrolment and post-V3 (Fig. S1B).A decrease in antibody titre between the post-V2 and pre-V3 timepoint was also observed in both liver disease and HC groups and was subsequently boosted by the administration of a third vaccine dose (Table S1).mRNA platforms were associated with significantly higher post-V2 anti-RBD titres compared to ChAdOx1 in both liver disease (953 [158-3,214] mRNA vs. 593 U/ml [114-1521] ChAdOx1; p = 0.0005) and HC cohorts (15,634 U/ml [10,829-21,445] mRNA vs. 1,198 U/ml [855-1,546] ChAdOx1; p < 0.001).Thirteen patients who received heterologous first and second vaccines had significant elevations in post-V2 antibody titres compared to homologous ChAdOx1 (p = 0.0002) and homologous mRNA regimens (p = 0.004) (Fig. 1B).In cohorts where data were available for both BNT162b2 or mRNA-1273 vaccinated individuals, there were no significant differences in anti-RBD at the post-V2 timepoint; therefore, both mRNA vaccines were grouped for further analysis (Fig. S2).Within the entire study cohort (patients and HCs) multivariable analyses showed that the factors significantly associated with lower antibody response after V2 were advancing age and inclusion in the LTR group, and factors associated with greater response were mRNA vaccination, heterologous vaccination (V1 + V2) and previous COVID-19 infection (Fig. 1C, Table S2).
Our study allowed us to compare antibody responses relating to specific combinations of disease and vaccine types.At the post-V2 and post-V3 timepoints LTRs mounted lower antibody titres compared to all other disease groups and HCs regardless of vaccine type (Table S3).mRNA-vaccinated patients with cirrhosis had reduced post-V2 antibody titres compared to mRNA-vaccinated HCs, but ChAdOx1-vaccinated patients with cirrhosis had comparable post-V2 titres to ChA-dOx1-vaccinated HCs.
We observed variable non-response rates across disease groups depending on vaccine type and timepoint (Table S4).No HCs were seronegative after either two or three vaccine doses.LTRs had the highest rates of serological non-response with 18/52 (35%) and 52/179 (29%) having absent responses after two doses of ChAdOx1 and mRNA vaccines, respectively.Non-response rates were reduced at the post-V3 timepoint in all disease groups, with only 9/97 (9%) LTRs and 2/108 (2%) patients with cirrhosis not having a serological response to vaccine at the post-V3 timepoint.

Serological cross-reactivity of SARS-CoV-2 VoC
IgG binding to the spike protein of SARS-CoV-2 Omicron subvariants was assessed at post-V2 and post-V3 timepoints in selected samples from liver disease and HC cohorts (Fig. 2A).Within the combined cohort (liver disease and HCs) at both timepoints, serological titres to all Omicron subvariants (except for BF.7 and BQ.1) were significantly lower compared to WT SARS-CoV-2.The subvariants with the greatest decrease in IgG binding compared to WT were BA2.75.2 (median fold decrease: x8.93 post-V2 and x6.12 post-V3) and BA.4.6 (x4.46 post-V2 and x3.2 post-V3) which were both first identified in autumn 2022.Notably, the magnitude of reduction in IgG binding to Omicron subvariants relative to WT was lower following V3 compared to post-V2.The same trends were observed when IgG binding was split by liver disease aetiology (Fig. 2B and Fig. S3A).However, HCs had less of a decrease in IgG binding to Omicron subvariants relative to WT than seen in patients with liver disease.Serum from most participants inhibited ACE2 binding to WT RBD at post-V2 and post-V3 (Fig. 2C), whereas there was a significant decrease in inhibition across all Omicron subvariants relative to WT.Again, the decrease in ACE2 binding post-V3 was less pronounced compared to post-V2.There was also less of a decrease at the post-V2 and post-V3 timepoints in HCs compared to patients with liver disease (Figs 2D and S3B).The ratio of IgG binding to WT spike compared to VoC spike was significantly increased by a third vaccine in the liver disease group, but no change was observed between two and three vaccine responses in HCs (Fig. S3C).We observed significant positive correlations at post-V2 and post-V3 timepoints when comparing the Roche anti-RBD titre with VoC binding IgG, ACE2 inhibition, and with the ratio between WT to VoC binding IgG (Fig. S4A-C).

Responses after COVID-19 vaccination
Liver transplant recipients Antibody responses to two and three vaccine doses in LTRs are presented in Fig. 3A and are separated according to class of immunosuppression.This shows a downward trend in post-V2 anti-RBD titres associated with increasing intensity of immunosuppression, with significant reductions observed in patients on a calcineurin inhibitor (CNI) plus mycophenolate mofetil (MMF) vs. a CNI alone, and in patients on a CNI plus another immunosuppressant other than MMF, compared to CNI alone.LTRs on MMF additionally had significantly reduced responses compared to those on thiopurines (p = 0.011, Fig. S5A), and an increasing daily dose of MMF was significantly associated with decreased anti-RBD titres at the post-V2 timepoint (p = 0.031) (Fig. S5B).A third vaccine dose significantly improved antibody responses across all groups, except with mTOR inhibitor monotherapy where cohort numbers were small (Fig. 3A).LTRs had high rates of antibody non-response, with non-response rates at the post-V2 timepoint of 4/12 (33%) ).Among LTRs, both univariable and multivariable analyses showed that the factors significantly associated with reduced odds of seropositivity after V2 were age (65-74 age group) and CNI plus MMF (Fig. 3B, Table S5).LTRs with previous SARS-CoV-2 infection were removed from the logistic regression, as all of these patients had a detectable response after V2.

Patients with autoimmune hepatitis
Patients with AIH had higher post-V2 antibody responses than LTRs with both ChAdOx1 and mRNA platforms despite both groups being immunosuppressed and being of similar age (61 years [49-69] vs. 60 years [52-68]).Unlike LTRs, there were no significant differences in serological response between class and intensity of immunosuppression at post-V2, despite the fact some patients with AIH (6/8 with MMF dose data) were on high dose (2 g/day) MMF (Fig. S6A).There were also no differences in response when the AIH cohort was split by the presence or absence of cirrhosis (Fig. S6B).Univariable and multivariable analyses showed that the factors significantly associated with higher antibody responses after V2 were mRNA vaccination and previous COVID-19 (Fig. 3C

Patients with cirrhosis
Antibody responses in patients with cirrhosis at post-V1, post-V2, and post-V3 are presented in Fig. 4A and are separated according to vaccination type and CP class.This revealed a dynamic interaction between number of doses, severity of cirrhosis, and vaccine type.At post-V1 there were no differences in antibody titres between compensated (CP-A) and decompensated (CP-B/C) cirrhosis when vaccinated with ChAdOx1.However, CP-A did have higher antibodies than CP-B/C at post-V1 when vaccinated with mRNA.At post-V1, both CP-A and CP-B/C had lower antibody titres than HC with both vaccine types.At post-V2, there were no significant differences between CP-A and CP-B/C or between patients and HCs when vaccinated with ChAdOx1.Whereas at post-V2 for mRNA, HCs had higher titres than patients with cirrhosis irrespective of CP class.At post-V3, there were no significant differences between any groups.Broadly, this suggested an association between liver disease severity and reduced antibody responses when vaccinated with mRNA but not the ChAdOx1 platform, particularly early in the vaccination course.To explore the interaction between disease severity and vaccine type further we plotted the correlation between MELD score and anti-RBD titres (Fig. 4B).At post-V1, this again showed that increasing disease severity was associated with decreased anti-RBD titres when vaccinated with mRNA platform but not with ChAdOx1.This trend persisted after V2 and V3, although it was non-significant.Among patients with cirrhosis, both univariable and multivariable analyses showed that age over 75 years and increasing MELD score were associated with a lower antibody response, whereas mRNA platform and previous COVID-19 were associated with higher titres (Fig. 4C, Table S7).

Patients with vascular liver disease
All patients with VLD were vaccinated with an mRNA platform.

T-cell responses
T-cell responses to WT SARS-CoV-2 across liver disease phenotypes T-cell IFNc responses to WT virus for a subset of infectionnaïve participants from each disease group are presented in Fig. 5A and Fig. S7.The majority of patients across the entire cohort had positive T-cell responses (>26 SFU/10 6 PBMCs) after at least one dose of vaccination.Within the liver disease cohort all patients generated a positive T-cell response after V2 except for 3/10 (30%) LTRs, 4/24 (17%) with cirrhosis, and 1/12 (8%) with AIH.Despite significant heterogeneity, all groups except for LTRs had a significant increase in the magnitude of IFNc responses after two or three vaccine doses (Fig. 5A).Within the total liver disease cohort, patients with previous COVID-19 had significantly higher IFNc responses after V1, V2, and V3 (Fig. 5B).There were positive correlations between post-V2 T-cell responses to WT spike, anti-RBD binding antibodies to WT, and functional antibody responses to all variants (ACE2 binding inhibition) (Figs 5C and S4A).

T-cell responses to Omicron BA.1 SARS-CoV-2 across liver disease phenotypes
To determine the cross-reactivity of vaccine-induced cellular responses we additionally assessed IFNc T-cell responses to peptides covering the Omicron (BA.1) spike protein (Fig. 5D).
Compared to the WT antigen, T-cell responses to BA.1 spike were well preserved regardless of disease group at post-V2 and post-V3 timepoints.However, when only assessing responses to peptides that differed between WT and BA.1 spike (mutated peptide pools; "minipools") there were significant reductions in BA.1 peptide-specific reactivity compared to WT in all groups (Fig. 5E), indicating that T-cell responses specifically to mutated epitopes were reduced but overall responses were maintained.5A).

Discussion
In a large, international, prospective study we assess humoral and cellular immune responses to multiple COVID-19 vaccine platforms across a range of liver disease types and severities using standardised timepoints and laboratory assays.We report on functional antibody and cellular responses to novel viral variants including a number of the most up-to-date Omicron subtypes, and assess COVID-19 infection rates and disease severity.
Longitudinal serum sampling demonstrated a stepwise increase in the magnitude of anti-SARS-CoV-2 RBD antibodies following one, two and three vaccine doses, and following past SARS-CoV-2 infection.In our cohort, use of the mRNA platform was associated with greater antibody titres compared to ChA-dOx1 which is in line with other datasets in healthy and immunosuppressed cohorts. 20,21In addition, as observed in healthy populations, 22 heterologous first and second vaccination was associated with >5-fold increase in post-V2 antibody responses compared to homologous vaccine delivery, suggesting that this approach should also be considered in liver cohorts.
Our data identifies LTRs as a particularly vulnerable cohort, having the lowest post-V2 antibody titres compared to all other disease groups.Although the number of cases of severe breakthrough infection in our cohort was small, it is notable that 4/5 (80%) were LTRs, two of whom received MMF therapy and had absent anti-RBD responses after V2.Vaccine immunogenicity in LTRs was heavily influenced by intensity of immunosuppressive medication with MMF associated with low antibody titres and high rates of non-response and MMF dose negatively associated with anti-RBD titres.This supports evidence from other patient groups that short-term discontinuation or dose reductions of immunosuppressive therapy may help maximise antibody responses to vaccination. 23,24Although other studies have reported suboptimal antibody responses in LTRs, 25,26 our cohort is notable for its size and geographic diversity.It also allows for direct comparisons with immunosuppressed patients with AIH who have more robust antibody responses despite being of similar age.This is most likely accounted for by the predominant use of thiopurines and corticosteroids, and the absence of MMF plus CNI dual therapy in patients with AIH.Reassuringly, irrespective of immunosuppressive status, antibody titres and rates of seroconversion were universally improved in all groups following the third vaccination, though some LTRs remained nonresponsive to vaccination.120 Journal of Hepatology, January 2024.vol.80 j 109-123

Responses after COVID-19 vaccination
In parallel with increasing serological titres to WT SARS-CoV-2, the cross-reactivity of vaccine-induced antibodies to VoC also improved across the liver disease cohort between post-V2 and post-V3 timepoints.This suggests that increasing the magnitude of serological response through a third vaccination maximises the likelihood that a proportion of antibodies will cross-react with VoC even in individuals who may have reduced capacity to produce high-affinity class-switched or somatically hypermutated antibodies. 27Despite improvements in antibody function with repeat vaccination, there was still a significant decrease in IgG binding and ACE-2 binding inhibition to nearly all Omicron subvariants in patients with liver disease after V2 and V3 relative to WT virus.This immune escape may partly account for the clear stepwise increments in breakthrough infection rates observed in our cohort.
Early in the vaccination course, severity of cirrhosis (indicated by CP class and MELD score) was associated with lower antibody responses to mRNA but not ChAdOx1.Although this effect was ultimately overcome by repeated vaccine doses it does point towards differential immunological mechanisms governing antibody response according to vaccine type.mRNA and adenoviral vector platforms are thought to induce antibody production through varying biological pathways which may be differentially impacted by cirrhosis-associated immune dysfunction. 28,29Further work is required to decipher the complex interplay between cirrhosis-associated immune dysfunction and immune responses to COVID-19 vaccination.Nonetheless, it is important to note that mRNA vaccines still induce higher antibody titres than ChAdOx1 at the post-V2 timepoint.
Despite heterogeneity in the magnitude of T-cell responses across HCs (as observed elsewhere 18 ) and disease cohorts, the majority of assessed individuals (88%) mounted an IFNc response to WT SARS-CoV-2 spike antigens after a single vaccine dose which was preserved after V2 and V3.1][32] Patients with VLD had particularly robust IFNc T-cell responses compared to other disease cohorts, which is possibly related to the absence of immunosuppressive medications, a possible unknown biological mechanism, preserved liver function, and use of mRNA-1273 vaccination which has previously been shown in healthy individuals to be associated with higher CD4+ and CD8+ SARS-CoV-2-specific T-cell responses. 21Overall, T-cell responses in individuals with liver diseases were also well maintained against the Omicron BA.1 variant, however reduced responses to BA.1 minipools suggest that responses to specific mutated epitopes may be lost. 33The cellular immune response to vaccination has emerged as a major determinant of individual risk of developing severe COVID-19, including in immunocompromised individuals. 13,14This may help explain why the majority of breakthrough infections reported in the liver disease cohort were mild-moderate with only 5/122 (4%) total infections reported as severe.However, post-V2 T-cell responses were detectable in the two participants with severe breakthrough COVID-19 and available PBMCs in this cohort, and other factors including potentially reduced virulence of later SARS-CoV-2 variants may additionally impact this. 34here are some limitations to our study.Firstly, despite recruiting from four different countries with diverse immunisation regimens, certain vaccine platforms in particular groups are lacking, including mRNA-1273-vaccinated HCs, third dose of ChAdOx1 in HCs, and ChAdOx1-or BNT162b2-vaccinated patients with VLD.Although it is tempting to extrapolate the immunological principles identified across three vaccines, the precise immune changes after multiple subsequent vaccine doses remain to be determined.Another constraint of our dataset is that the HC cohort is comprised of healthcare workers who are significantly younger with fewer comorbidities than the liver disease population.However, we have performed a multivariable analysis of the entire cohort which accounts for age in order to identify cofactors and disease groups associated with vaccine response.The breakthrough COVID-19 data must also be interpreted with caution as asymptomatic infection may not have been identified and it remains impossible to fully account for important confounding variables such as local SARS-CoV-2 prevalence, viral load exposure, further vaccine doses, and individual patient behaviours including shielding measures.Furthermore, we were unable to systematically collect accurate data on the use of antiviral medications and recombinant antibodies due to incomplete documentation in electronic hospital records and geographic variability in access to these agents.As a result, we have opted to remain descriptive with this domain of the study and have not performed statistical analyses.Lastly, due to sampling limitations, IFNc T-cell assay results were not available in all participants with severe breakthrough infection.
In summary, we demonstrate that the three most widely available vaccine platforms are immunogenic and appear to protect against severe COVID-19 in a diverse group of patients with a variety of underlying liver conditions.Even patients with advanced cirrhosis mount robust immune responses after two and three vaccine doses irrespective of vaccine type.This will provide reassurance to patients with chronic liver disease who were previously deemed at high risk of severe COVID-19 and death during the pre-vaccination era.In addition, our data will be encouraging in the event of future unforeseen viral infections which may also require rapid vaccine development and delivery to patients with liver disease.However, we show that LTRs mount lower antibody and T-cell responses, related to intensity of immunosuppression and the use of MMF, with most cases of severe COVID-19 occurring in this patient group.We recommend that LTRs should be vigilantly monitored for the development of severe COVID-19 if infected, and prioritised for repeated vaccination, prophylactic antiviral agents, and enrolment into trials exploring the role of immunosuppressive dose modification and alternative vaccine strategies.

Breakthrough SARS-CoV-2 infection after COVID-19 vaccination
Rates of breakthrough infection were plotted over time for each of the 4 recruiting countries alongside the corresponding proportions of circulating viral variants.Country-specific proportions of SARS-CoV- log 10 anti-RBD estimate (95% CI)

Fig. S1 .
Fig. S1.A) Magnitude of anti SARS-CoV-2 RBD Ig in infection naïve and previously SARS-CoV-2 infected individuals at post-V1 and post-V2 timepoints.B) Magnitude of anti SARS-CoV-2 RBD Ig in Naïve individuals at post-V2 and post-V3 timepoints and in individuals who became nucleocapsid positive between second and third vaccines (Post-V3 Newly N+ve).Boxes represent median and IQR, whiskers represent +/-1.5xIQR.Mann Whitney U test used, adjusted P value presented.

Table S2 :
Linear regression model of post-v2 log10 transformed anti-RBD Ig across entire cohort.Age variable is compared to 18-44year old age group.* indicates significant values

Table S3 :
In infection naïve individuals, comparison of Roche anti-RBD antibody response across disease groups at post-V2 and post-V3.Comparisons at post-v3 in AstraZeneca vaccinated individuals not made due to low n numbers.Kruskal Wallis with Dunn's post-hoc test, adjusted for multiple comparisons using Benjamini Hochberg.Cir = Cirrhosis, AIH = Autoimmune hepatitis, LT = Liver transplant, Az = AstraZeneca vaccine.* indicates statistical significance (P<0.05)

Table S7 :
Linear regression models of log10 transformed anti-RBD antibody (>0.8U/mL) in cirrhosis patients at the Post-V2 timepoint.MELD is a continuous variable, all other variables are discrete.Age is compared to 18-44 year old group.CP = Child's Pugh class, INR = International normalized ratio, CI = Confidence interval.