Characteristics of an antibody profile associated with protection against hepatitis C virus reinfection in a high-risk cohort

Abstract

Background

Without a vaccine, HCV (re)infections continue to afflict individuals at high risk, for example, men who have sex with men (MSM).

Objective

To guide vaccine development, we assessed whether natural protection against observed HCV reinfection among MSM was associated with a distinct humoral immune profile.

Design

From a prospective cohort of HIV-positive MSM with acute HCV infection (MOSAIC, Dutch MSM Observational Study of Acute Infection with hepatitis C), we selected 13 cases who became HCV-reinfected (R) and 18 controls who remained free of observed reinfections (non-reinfected, NR) in 0.5–9.5 years after primary infection clearance, matched for calendar year of primary infection. At time point T₁ (29 days before to 78 days after primary clearance) and T₂ (3–6 months before R subjects’ first reinfection), we evaluated serum antibody responses against HCV’s E1E2 envelope glycoprotein. Based solely on E1E2-binding antibody levels and Fc-gamma receptor (FcγR)/C1q engagement at T₁, we divided participants into subgroups using latent profile analysis.

Results

HCV-neutralising responses were observed mainly at T₁ (9/18 NR and 3/13 R participants). The NR group could be distinguished from the R group at T₁ by higher levels of E1E2-binding antibodies that bound FcγRs/C1q. Among four identified latent profiles, HCV reinfection prevalence was lowest in the profile defined by moderate to high levels of E1E2-binding antibodies (IgM/IgG/IgG1/IgG3) and FcγR/C1q engagement. This profile exhibited a relatively broad, potent HCV-neutralising capacity and the broadest epitope specificity.

Conclusion

We propose broad, potent serum Fc-mediated effector functions and neutralisation and a broad anti-E1E2 antibody epitope specificity as ingredients of HCV-protective immunity towards which vaccination approaches should aim.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• This study uncovered a serum anti-E1E2 antibody profile around the time of primary HCV clearance that was associated with a lack of HCV reinfection in a high-risk cohort of men who have sex with men. Our results suggest that the mechanistic explanation for an HCV reinfection-protective phenotype may potentially involve the elicitation of a primary anti-E1E2 antibody response characterised by broad, potent Fc-mediated effector functions and neutralisation and a broad epitope specificity.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The primary antibody response profile associated with HCV reinfection protection may represent the antibody response that experimental vaccines should induce to achieve protective immunity.

Introduction

In 2022, ~0.98 million new HCV infections occurred globally.¹ Whereas 20–30% of people spontaneously clear HCV, 70–80% develop chronic infections.² Chronic HCV infection affects ~50 million people worldwide and is a major cause of life-threatening liver-related complications.¹ Direct-acting antivirals can effectively cure HCV, but both treatment-induced and spontaneous clearance leave individuals susceptible to reinfection on re-exposure. Reported reinfection rates in populations at high risk of HCV acquisition, for example, people who inject drugs and men who have sex with men (MSM), range from 0.0 to 31.0/100 person-years.³ This reinfection risk underscores the need for a prophylactic vaccine to meet the WHO’s goal of eliminating HCV as a public health threat by 2030.^{4 5}

Vaccine development can be guided by defining immunological correlates of protection against HCV persistence and reinfection. Research has established a major role for T cell immunity in (re)clearance (reviewed in⁶). Additionally, anti-HCV antibodies may facilitate spontaneous clearance of primary infection via virion neutralisation by binding to HCV’s E1E2 envelope glycoprotein.^7–18 The early appearance in serum of broadly neutralising antibodies (bNAbs) recognising diverse E1E2s is associated with spontaneous resolution of acute infection^15–17 and passively transferred bNAbs can terminate infection and protect against HCV challenge in animal models.^{7–14 18 19} However, the role of neutralisation breadth and potency in preventing (persistent) reinfection remains unclear; one study found no association between reinfection outcome and serum neutralising capacity measured at one moment before or during reinfection,²⁰ while a different longitudinal analysis linked earlier NAb responses and E2-specific memory B cell expansion to spontaneous reclearance.²¹

Also relatively understudied are potential protective contributions from antibody-mediated effector functions, for example, antibody-dependent cellular phagocytosis (ADCP), antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent complement deposition (ADCD). To trigger these mechanisms (reviewed in^{22 23}), crystallisable fragment (Fc) tails of infected cell-bound or virion-bound antibodies initiate (sub)class-dependent interactions with effector cell Fc receptors (in the case of ADCP and ADCC) or with the C1q complement protein (in the case of ADCD). Subsequently, the effector cells phagocytose the infected cells or virions (ADCP) or kill them (ADCC), or complement activation (ADCD) promotes inflammation and phagocytosis or lysis of the infected cells. One or more of these Fc-mediated effector functions have been implicated in protective immunity against, for instance, influenza virus,^24–26 severe acute respiratory syndrome coronavirus 2,^{27 28} respiratory syncytial virus^{29 30} and HIV.³¹ However, Fc effector functions seem redundant for spontaneous clearance of the HCV-related rodent hepacivirus, since mice lacking the requisite Fc receptors or complement cleared infection normally.³² For HCV, higher serum ADCP activity was associated with higher anti-E1E2 antibody levels during early infection in individuals who developed persistent infection versus spontaneous resolvers.³³ Conversely, similar ADCC degrees have been reported across patients with acute, chronic or resolved HCV infection, despite higher E1E2-binding antibody levels during chronicity.³⁴

A key knowledge gap in protective humoral immunity against HCV is whether antibody-mediated neutralisation and Fc effector functions protect against HCV reinfection following primary infection clearance. Addressing this question requires comparing individuals who remain reinfection-free despite high-risk behaviour to individuals who become reinfected. The Dutch MSM Observational Study of Acute Infection with hepatitis C (MOSAIC)³⁵ provides a valuable resource for such research, as it followed an HIV-positive cohort with a high HCV reinfection incidence (11.5/100 person-years).³⁶ With regularly collected serum samples and detailed behavioural and reinfection monitoring,³⁷ MOSAIC offers a unique opportunity to explore the relationship between humoral HCV immune responses and HCV reinfection.

Here, we evaluated anti-HCV antibody responses in 13 HCV-reinfected (R) and 18 non-HCV-reinfected (NR) MOSAIC participants with spontaneous or treatment-induced clearance of primary infection. Our findings offer new insights into components of HCV-protective immunity that vaccination approaches should aim to elicit.

Patients and methods

Patients

Patients were selected from MOSAIC, a prospective cohort study of HIV-positive MSM with acute HCV infection conducted between 2009 and 2017 in the Netherlands.³⁵ Sera and clinical, virological and HCV risk behaviour data were collected at cohort entry and follow-up study visits (median between-visit interval: 56 days; IQR: 28–99). Primary HCV infection was defined as transitioning from negative to positive anti-HCV antibody status (AxSym HCV V.3.0; Abbott) and to positive HCV RNA status (Versant; Siemens Medical Solutions Diagnostics); the midpoint between the date of the last negative and first positive test represented the estimated start of primary infection. HCV infection clearance was defined as a sustained virological response (HCV RNA-negative 12 or 24 weeks after completing treatment), or spontaneous clearance (anti-HCV antibody-positive and HCV RNA-negative on two consecutive occasions within 1–2 months without treatment); the midpoint between the last HCV RNA-positive and subsequent HCV RNA-negative test represented the estimated clearance date.

Participants were classified as R or NR based respectively on detected reappearance or continued absence of HCV RNA during ≥0.5 follow-up years after primary infection clearance. For this study, we included all R participants (13) and NR participants (18) who could be matched on calendar year of primary HCV infection. We assessed R and NR participants at two time points: T₁ (29 days before to 78 days after the estimated clearance date) and T₂ (3–6 months before R participants’ first reinfection or an equivalent follow-up time point for NR participants, provided T₂ occurred ≥5 months post-T₁ and HCV RNA was negative for ≥2 measurements between T₁ and T₂).

Additionally, the following samples were used for binding assays: pooled sera from 4 R participants collected 6 months before primary infection (negative control); 10 healthy donor sera (negative controls) and serum from a chronically HCV-infected MOSAIC participant (positive control).

Methods

Please consult the online supplemental methods, online supplemental figure 1 for a description of biological measurements and statistical analyses.

Results

Characteristics of study participants

We included 31 MOSAIC participants who cleared a primary HCV infection within 3.3–15.1 months, either spontaneously (n=4) or via treatment (n=27, 25 of whom received interferon with(out) ribavirin; table 1, online supplemental tables 1,2). All participants were living with and treated for HIV and most (27/31) were of Western/Southern European ethnicity. During 0.5–9.5 follow-up years, 18/31 participants remained free of observed reinfection (NR), while 13/31 became R one to four times (median: two reinfections, IQR: 1–2; online supplemental figure 2) with the first reinfection occurring a median of 2.0 years (IQR: 1.1–2.7) after primary infection clearance.

Table 1. Characteristics of study participants, stratified by HCV reinfection occurrence.

				Total (n=31)		Non-reinfected (NR) (n=18)		Reinfected (R) (n=13)		P value
Ethnicity										0.899
	Western/Southern European			27 (87.1)		15 (83.3)		12 (92.3)
	African			1 (3.2)		1 (5.6)		0 (0.0)
	Asian			1 (3.2)		0 (0.0)		1 (7.7)
	North American			1 (3.2)		1 (5.6)		0 (0.0)
	Surinamese			1 (3.2)		1 (5.6)		0 (0.0)
Interval between last HCV-negative test before and first HCV-positive test during primary HCV infection, months*				5.5 (3.4–6.6)		5.5 (3.4–6.4)		5.3 (3.3–7.7)		0.818
Calendar year of primary HCV infection				2009 (2008–2010)		2009 (2008–2011)		2008 (2007–2010)
Age at start of primary HCV infection, years				41.6 (36.6–48.5)		45.6 (40.7–49.9)		36.9 (35.6–41.4)		0.045
CD4+ cell count at start of primary HCV infection, cells/mm3				510 (410–600)		505 (378–555)		510 (440–670)		0.575
Subtype of primary HCV infection										0.214
	1a			14 (45.2)		9 (50.0)		5 (38.5)
	1b			2 (6.5)		2 (11.1)		0 (0.0)
	2b			2 (6.5)		2 (11.1)		0 (0.0)
	3a			1 (3.2)		1 (5.6)		0 (0.0)
	4d			8 (25.8)		3 (16.7)		5 (38.5)
	Unknown			4 (12.9)		1 (5.6)		3 (23.1)
Duration of primary HCV infection, months				7.1 (5.4–8.4)		7.0 (5.6–8.1)		7.3 (5.3–8.6)		0.704
Interval between estimated start of primary HCV infection and time point T1, months				6.5 (5.3–7.7)		6.2 (5.2–7.4)		6.7 (5.5–8.1)		0.719
Interval between last HCV-positive test during and first HCV-negative test after primary HCV infection, months†				0.8 (0.5–1.6)		0.8 (0.5–1.5)		1.1 (0.6–1.7)		0.525
Interval between time point T1 and estimated date of primary HCV clearance, days				12 (7–24)		13 (8–23)		11 (7–23)		0.534
Mechanism of primary HCV clearance										0.018
	Spontaneous			4 (12.9)		0 (0.0)		4 (30.8)
	Treatment-induced			27 (87.1)		18 (100.0)		9 (69.2)
			IFN with(out) RBV		25 (92.6)		16 (88.9)		9 (100.0)
			DAA		2 (7.4)		2 (11.1)		0 (0.0)
Follow-up from estimated date of primary HCV clearance until last HCV RNA test, years				5.2 (3.1–6.0)		4.6 (2.7–5.6)		5.9 (5.2–7.4)		0.060
Interval between time points T1 and T2, years‡				1.9 (0.9–2.7)		1.9 (0.9–2.8)		2.0 (1.0–2.5)		0.963
Interval between time point T2 and first HCV reinfection, days‡						NA		84 (62–93)
Genotype switches between primary HCV infection and first HCV reinfection						NA
		1a>2 c						1 (7.7)
		1a>4d						1 (7.7)
		4d>1a						5 (38.5)
		No switch: 1a>1a						2 (15.4)
		Unknown						4 (30.8)
Number of HCV reinfections				0 (0–1)		NA		2 (1–2)		<0.001

All participants were living with HIV and participated in the Men who have sex with men Observational Study of Acute Infection with hepatitis C. Reported values are medians (IQR) for numerical variables and counts (%) for categorical variables. P values for comparisons between non-reinfected and reinfected individuals were calculated from Kruskal-Wallis tests for numerical variables and Pearson’s χ² tests for categorical variables.

^*Missing data for one non-reinfected participant.

^†Missing data for one reinfected participant.

^‡No T₂ sample available for one reinfected case and two matched non-reinfected controls.

DAA, direct-acting antiviral(s); IFN, interferon; NA, not applicable; RBV, ribavirin; T₁, 29 days before to 78 days after primary clearance; T₂, 3–6 months before R subjects’ first reinfection.

No significant differences were noted between the NR and R group in CD4⁺ cell counts at the estimated start of primary infection (p=0.575). However, NR participants were older at their first HCV infection (p=0.045). Among 27 participants with a genotyped primary infection, subtypes 1a (9/18 NR individuals; 5/13 R individuals) and 4d (3/18 NR individuals; 5/13 R individuals) predominated. Notably, in 7/9 R individuals with a genotyped first reinfection, the subtype of reinfection and primary infection differed.

Limited neutralising capacity in sera from R participants around the time of primary infection clearance

To evaluate the role of neutralising antibody responses in remaining reinfection-free, we tested polyclonal antibodies (pAbs) purified from NR and R sera from a time point near primary clearance (T₁: 29 days before to 78 days after; table 1, online supplemental tables 1,2) against a 12-HCV pseudoparticle (HCVpp) panel representing genotypes 1–6 and various neutralisation sensitivities³⁸ (figure 1). The NR group tended to exhibit stronger neutralising responses, as reflected by a numerically higher geometric mean titre (GMT) and breadth (number of HCVpps neutralised with a 50% inhibitory dilution (ID₅₀) >20) versus the R group. However, these differences lacked statistical significance (p=0.085 and p=0.073 for the GMT and breadth, respectively) and there was considerable interindividual variation in neutralising responses among NR individuals: 9/18 NR individuals neutralised 1–11/12 HCVpps (with ID₅₀s ranging from 27 for AMS0230, the most resistant strain, to ~5800 for UKNP5.2.1, the most sensitive strain), while serum pAbs from the remaining 9 NR participants were non-neutralising at the 1:20 starting dilution. Neutralisation in the R group was limited to 3/13 individuals whose sera neutralised 1–5 HCVpps (ID₅₀s: 27–457).

Figure 1. T₁ serum HCV pseudoparticle (HCVpp)-neutralising capacity of HCV-reinfected (R) and non-reinfected (NR) individuals. Left panels: 50% inhibitory dilutions (ID₅₀s) for serum polyclonal antibody (pAb)-mediated neutralisation of HCVpps from a panel encompassing four neutralisation sensitivity tiers and six genotypes (GTs). Middle panels: geometric mean titres (GMTs) calculated from ID₅₀s of tested HCVpps. Right panels: numbers of HCVpps neutralised by ≥50% at the starting serum pAb dilution (1:20). The GMT and breadth were numerically yet non-significantly different between the R and NR group (Mann-Whitney: p=0.085 and p=0.073, respectively). T₁, 29 days before to 78 days after primary clearance.

We also tested serum pAbs from T₂ (3–6 months before R individuals’ first reinfection or a comparable time point for matched NR individuals, if available; table 1) against the seven most sensitive HCVpps. Most T₂ serum pAbs were non-neutralising (online supplemental figure 3). Among eight NR individuals whose serum pAbs neutralised ≥1 HCVpp at T₁ and for whom T₂ sera were available, neutralisation had universally declined, with T₂ serum pAbs neutralising ≤1 HCVpp. Notably, increased potencies against four or one HCVpp were measured for two R individuals, possibly reflecting post-clearance HCV exposure insufficient to establish infection. Overall, serum neutralisation capacities did not clearly distinguish NR from R individuals 3–6 months before reinfection.

Stronger engagement of E1E2-specific antibodies with activating Fc receptors and C1q in sera from NR subjects around the time of primary infection clearance

Since Fc effector functions are linked to protective immunity against various viruses, we examined whether remaining HCV reinfection-free was associated with a higher potential for activating such functions. Using a Luminex multiplex assay with 11 different membrane-bound E1E2 proteins (mbE1E2s)³⁸ (online supplemental methods), we measured levels of E1E2-specific serum antibodies engaging with activating Fc-gamma receptors (FcγRIIA/FcγRIIIA, both involved in ADCP and ADCC) or C1q (involved in ADCD)^{22 23} (figure 2, online supplemental figure 4). Bayesian linear regression revealed significantly stronger engagement with FcγRIIA, FcγRIIIA and C1q in T₁ serum of NR versus R participants for 8/11, 6/11 and 7/11 mE1E2s, respectively (figure 2). At T₂, significant differences were lost, mainly due to waning FcγR/C1q engagement among NR individuals (online supplemental figure 4). For most R participants, FcγR/C1q engagement at T₁ and T₂ approximated background signals in negative control sera.

Figure 2. T₁ serum levels of HCV E1E2-binding antibodies engaging with Fc-gamma receptors (FcγRs) and C1q among HCV-reinfected (R) and non-reinfected (NR) individuals. Heatmaps: antibody levels per individual. Positive controls: cHCV, chronically HCV-infected individual; RSV F, respiratory syncytial virus soluble fusion protein. Negative controls: HD, 10 healthy donors (median); T₋₁, pre-infection pool (four individuals). MFI, median fluorescence intensity; mbE1E2, membrane-bound E1E2 protein. Dot plots: Bayesian linear regression of differences in mean antibody levels (Δs) between the NR (set at 0 (red line)) and R group. Points with bands represent medians of Δ’s posterior distribution with associated 95% credible intervals (CrIs). T₁, 29 days before to 78 days after primary clearance.

Stronger engagement of E1E2-specific antibodies with activating Fc receptors and C1q is paralleled by increased E1E2-binding IgG in NR subjects

Having observed serum-mediated HCVpp neutralisation for 9/18 NR individuals and greater FcγR/C1q engagement in NR versus R sera around the time of primary clearance, we investigated whether these features corresponded to differences in E1E2-binding antibody (sub)class levels. Exploiting a Luminex assay with aforementioned mbE1E2s, we measured E1E2 binding by IgM, IgG (including IgG1, IgG2, IgG3 and IgG4) and IgA (figure 3, online supplemental figures 5,6). All these antibody (sub)classes can neutralise virions.³⁹ Bar IgG4, these antibodies may also activate complement and all IgG subclasses can engage FcγRs.³⁹

Figure 3. T₁ serum levels of HCV E1E2-binding IgM, IgG, IgG1 and IgG3 among HCV-reinfected (R) and non-reinfected (NR) individuals. Heatmaps: antibody levels per individual. Positive controls: cHCV, chronically HCV-infected individual, RSV F, respiratory syncytial virus soluble fusion protein. Negative controls: HD, 10 healthy donors (median); T₋₁, pre-infection pool (four individuals). MFI, median fluorescence intensity; mbE1E2, membrane-bound E1E2 protein. Dot plots: Bayesian linear regression of differences in mean antibody levels (Δs) between the NR (set at 0 (red line)) and R group. Points with bands represent medians of Δ’s posterior distribution with associated 95% credible intervals (CrIs). T₁, 29 days before to 78 days after primary clearance.

At T₁, NR participants had significantly higher levels of IgG binding to 7/11 mbE1E2s (figure 3). This difference was primarily IgG1-driven and, to a lesser extent, IgG3-driven, with IgG1 and IgG3 levels in NR sera significantly exceeding their counterparts in R sera for 3/7 and 1/7 mbE1E2s, respectively. By T₂, most significant differences in IgG/IgG1/IgG3 binding between the NR and R group had diminished, likely due to waning antibody levels in NR individuals (online supplemental figure 5). Binding IgM levels were comparable between groups at T₁ and T₂, except for a significantly higher H77-binding level in T₂ NR sera. In contrast, IgA/IgG2/IgG4 levels at T₁ and T₂ in both groups were generally indistinguishable from negative control background signals (online supplemental figure 6). These findings suggest that differences in neutralising capacity and FcγR/C1q engagement between NR and R subjects were most closely linked to E1E2-binding IgG, particularly IgG1/IgG3.

To further evaluate the relation between T₁ binding antibody levels and neutralising potency or FcγR/C1q engagement, we calculated the Bayesian Pearson’s correlation coefficient r for these variables per HCV strain in each participant group, focusing on antibodies with levels above background (online supplemental figure 7). In the NR group, moderate to strong positive correlations (r>0.50) with strong to extreme plausibility (Bayes Factor ₁₀(BF₁₀) ≥10) were observed particularly between IgG levels and FcγRIIA/IIIA interaction and between IgG3 levels and C1q interaction. Conversely, correlations in the R group were generally weak or lacked strong evidence (BF₁₀<10), bar some correlations between IgG and FcγRIIA/IIIA interaction. Analysing correlations between binding and neutralisation, we found strong to extreme support only in the NR group for moderate to strong positive correlations between E1E2-binding IgG/IgG1 and HCVpp neutralisation for four strains (H77, AMS0232, UKNP4.1.1, UKNP3.2.2). In summary, especially the stronger engagement of NR serum E1E2-binding antibodies with FcγRIIA/FcγRIIIA/C1q at the time of primary clearance may be largely driven by elevated E1E2-binding IgG levels in NR individuals.

Latent profile analysis reveals components of an anti-HCV antibody profile associated with reinfection protection

Building on the above, we investigated the link between anti-HCV antibody responses and reinfection using latent profile analysis (LPA). LPA groups individuals into homogeneous subgroups (that is, latent profiles (LPs)) based on observed continuous indicators.⁴⁰ To reduce model complexity, we selected T₁ levels of FcγR-interacting and C1q-interacting antibodies and antibody (sub)classes binding to a subset of mbE1E2 variants as indicators. For these variants (H77, AMS0230, UKNP2.2.1, UKNP2.4.1, AMS.3a.k26, UKNP3.2.2, AMS.4d.k9, UKNP6.1.2), ≥3 mean antibody levels differed significantly (p<0.05) between the NR and R group (figures 2 and 3). Among models with one to five LPs, the 4-profile model was chosen for further analysis given its lowest Bayesian Information Criterion, indicating the most optimal goodness-of-fit in face of model complexity (online supplemental table 3). This model had an acceptable minimum proportion of individuals per LP (19.4%) and high classification reliability (entropy and average posterior membership probability of 1.00; online supplemental table 3). Sociodemographic or HCV infection characteristics did not differ significantly between LPs (online supplemental table 4).

Mean antibody levels per LP are shown in figure 4A and online supplemental figure 8. LPs were primarily distinguished by variables that also separated the NR and R group in the direct comparisons, that is, serum levels of mbE1E2-binding IgG/IgG1/IgG3 (figure 3) and antibodies engaging FcγR/C1q (figure 2), as well as IgM. LP4 (7/31 participants, all NR) exhibited the strongest overall binding, with high IgG, moderate IgM/IgG1/IgG3 and low IgG2/IgG4/IgA levels, alongside moderate FcγR/C1q interaction.

Figure 4. Levels of serum E1E2-binding antibody (Ab) at T₁ and prevalence of HCV reinfection across latent profiles (LPs) of anti-HCV Ab responses. (A) Mean serum levels (coloured dots) of membrane-bound (mb)E1E2-binding Ab (sub)classes or Abs interacting with Fc-gamma receptor (FcγR) ectodomains and C1q, per LP. Black points with error bars reflect global median Ab levels with associated SD. MFI, median fluorescence intensity. (B) Proportion of HCV-reinfected individuals per LP (points) and associated 95% credible intervals (CrIs) (bands). The OR and associated 95% CrI reflect the relation between increasing LP number and reinfection prevalence. T₁, 29 days before to 78 days after primary clearance.

To assess the HCV reinfection probability across LPs, we analysed the relationship between LP and the proportion of R individuals using Bayesian logistic regression (figure 4B). Reinfection prevalence ranged from>50% in LPs with generally the lowest antibody binding and FcγR/C1q interaction (LP1/LP2) to 33.3% (95% credible interval (CrI) 7.7% to 71.4%) in LP3, which had generally the second-highest binding and interaction levels, and 0% (95% CrI 0% to 29.2%) in LP4, which exhibited generally the highest levels. The OR of 0.32 (95% CrI 0.11 to 0.74) indicated that higher LP designations were associated with a greater probability of remaining reinfection-free. Thus, a combination of moderate to high serum E1E2-binding IgM, IgG (particularly IgG1/IgG3) and FcγR/C1q-interacting antibody levels around primary clearance was linked to HCV reinfection protection.

We further characterised the LPs by comparing HCVpp-neutralising capacity among LP1-4 individuals at T₁ (figure 5A, online supplemental figure 9). Neutralising breadth and GMT were highest in LP3 and LP4, which had the strongest antibody binding and lowest HCV reinfection prevalence (median breadth: 4 and 5; median GMT: 37 and 32, respectively; figure 5A). Both metrics were significantly higher (p<0.05) in LP4 versus LP1/LP2, which had the weakest antibody binding and highest reinfection prevalence (median neutralisation breadth and GMT: 0 and 20 for both). Notably, LP3 and LP4 included the 2 individuals with the greatest neutralising breadth (11/12 for both) and GMT (181 for the LP3 individual and 316 for the LP4 individual; online supplemental figure 9). These findings indicate that broad, potent E1E2 binding profiles, associated with HCV reinfection protection, were also associated with broad, potent neutralisation around the time of primary clearance.

Figure 5. T₁ serum HCV-neutralising capacity and antibody epitope specificity across latent profiles (LPs) of anti-HCV antibody responses. (A) Serum neutralising breadth (left) and potency (right). GMT, geometric mean 50% neutralisation titre; ID₅₀, 50% inhibitory dilution. (B) Competition Luminex assay. The binding signal of un-competed monoclonal antibodies (mAbs; with epitopes in brackets) to membrane-bound E1E2 proteins (mbE1E2s) was set at 100% (blue line) to calculate residual mAb-mbE1E2 binding after mbE1E2-serum preincubation. Each dot in (A, B) represents one participant’s serum. P values are indicated for significant differences in global medians between LPs (adjusted p<0.05; Bonferroni-corrected Kruskal-Wallis test). T₁, 29 days before to 78 days after primary clearance.

Finally, we employed a Luminex competition assay with the mbE1E2 panel to evaluate anti-E1E2 antibody epitope specificities across LPs (figure 5B, online supplemental figure 10). Specificities to bNAb epitopes (AR4, AS412, AR3, domain C/AR2 and E1’s stem) and the non-NAb domain A epitope were assessed by testing competition between nine reference mAbs and T₁ sera for binding mbE1E2s. Four mbE1E2-mAb combinations (UKNP2.2.1 with AP33, AMS.4d.k9 with AP33, UKNP6.1.2 with CBH-4B, AMS0230 with HC84.26) did not pass the binding threshold and were excluded. Analysing mean residual binding per individual per mAb (figure 5B) revealed that competition (<100% residual mAb binding) across all epitopes was primarily observed in LP3/LP4, which had the highest binding levels and lowest reinfection prevalence. Although differences between LP3 and LP4 were statistically non-significant (adjusted p for multiple comparisons >0.05), only LP4 showed median residual binding <90% for all mAbs. Moreover, LP4 displayed significantly lower median residual binding for AS412-targeting AP33 and AR4-targeting AR4A and AT1618 (adjusted p<0.05) versus LP1/LP2. Interestingly, binding of AP33 and E1 stem-targeting IGH505 was modestly enhanced by LP1/LP2 sera (median residual binding: 113–117%), as was binding of domain A-targeting CBH-4B by LP1 sera (median: 115%). Taken together, the profile with the highest E1E2-binding antibody levels and lowest HCV reinfection prevalence was uniquely characterised by multi-epitope competition between serum antibodies and mAbs for E1E2 binding, particularly at the AS412 and AR4 epitopes.

Discussion

This study aimed to identify an anti-HCV antibody profile associated with reinfection protection in an MSM cohort with a high reinfection rate.³⁶ Unlike prior studies that analysed factors underlying spontaneous clearance of a single infection, our prospective, longitudinal approach evaluated antibody responses that may prevent reinfection under real-world exposure conditions, treating primary infection as a ‘model’ for immunisation.

By comparing 18 NR with 13 R participants, we identified potential distinguishing features of protective immunity. Neutralising antibody responses were seen in 50% of NR participants, generally characterised by broad, potent activity, whereas only 1/13 R participants had such a response. Furthermore, significantly higher E1E2-binding antibody levels and enhanced interactions with FcγRIIA, FcγRIIIA and C1q were observed among NR participants. By constructing profiles based on E1E2-binding antibody levels and FcγR/C1q engagement, we found that the profile defined by the strongest binding and FcγR/C1q engagement was associated with the lowest reinfection risk and a relatively broadly and potently neutralising antibody response targeting various epitopes. The described profiles may underestimate anti-E1E2 IgG1 responses in some participants, as the anti-IgG1 antibody (4E3) used here was recently reported to preferentially recognise IgG1’s G1m17 allotype versus IgG1’s G1m3 allotype.^{41 42} This allotype specificity may explain the discrepancy between high total IgG levels and low IgG subclass levels observed particularly in LP3.

Perhaps also counterintuitively, reinfections were observed in all four participants who had spontaneously cleared a primary infection. It is tempting to speculate that their apparent lack of protection against reinfection is attributable to a lack of strong antibody responses around the time of primary clearance. Indeed, none of these four spontaneous clearers were assigned to the profile with the strongest anti-HCV antibody response (LP4) and two of them to the profile with the weakest anti-HCV antibody response (LP1). However, the small sample size of this study precludes drawing any meaningful conclusion from this finding. Moreover, in this cohort, treatment was generally initiated relatively quickly after HCV acquisition, without awaiting spontaneous clearance. As a result, spontaneous clearers may be under-reported and comparisons between participants who cleared spontaneously and participants with treatment-induced clearance are likely to be biased due to misclassification.

Previous studies involving spontaneous HCV clearers and chronic progressors,^7–18 or passively immunised animals,^{7–14 18 19} have indicated that a broad serum neutralising response is probably critical for vaccine efficacy. The findings presented here corroborate and extend this view by suggesting that both broad anti-HCV neutralising responses and Fc-mediated effector functions might protect against reinfection. However, Fc-mediated effector functions are regulated by complex integration of stimulatory and inhibitory signalling, involving more interactions than studied here.^{22 23} Therefore, it remains unclear whether the observed interactions between NR serum, E1E2 and FcγRIIA/IIIA or C1q effectuate enhanced phagocytosis or killing of HCV-infected cells and virions. Useful for quantifying relative contributions of Fc effector functions and neutralisation to in vivo protection could be FcγR-deficient/complement-deficient animal models, along with mAb derivatives incapable of FcγR/C1q engagement. Such approaches have been highly informative for other viruses.^{24 26 28} Should ADCP and ADCC indeed contribute to reinfection protection, this finding would contrast with previous reports of increased ADCP activity in chronic progressors versus spontaneous resolvers³³ and similar ADCC activity across acute, chronic and cleared infections.³⁴ Further investigation into the breadth of ADCP and ADCC responses is needed to clarify their role in HCV protection and relevance for vaccine development.

Additional insights for HCV vaccine design can be obtained from serum and mAb E1E2 binding competition experiments. Interestingly, only sera from the reinfection-free LP4 competed effectively with antibodies targeting bNAb epitopes AR4 and AS412. This result contrasts with earlier studies suggesting no significant differences in the relative dominance of antibody responses toward specific epitopes between spontaneous clearers and chronic progressors.^{20 43} Our data highlight AR4 and AS412 as possibly critical epitopes for eliciting protective immunity. AR4, a conformational epitope, and AS412, a linear epitope, are highly conserved across HCV strains,^{8 44 45} making them attractive targets for vaccine development. So far, several AS412-based vaccine candidates have been tested for their immunogenicity in mice, with varying success.^46–50 Our results encourage further evaluation of top candidates and novel E1E2 designs that effectively focus humoral immune responses on both AS412 and AR4A.

While the above discussion features circulating antibodies as likely contributors to reinfection protection, additional factors are presumably involved. Generally, long-term protection against secondary infections is mediated by T and B cell memory, the latter comprising long-lived plasma cells (LLPCs) and memory B cells. Whereas LLPCs continuously secrete antibodies, memory B cells rapidly differentiate into antibody-secreting cells on antigen re-exposure.⁵¹ LLPCs appear to have played a limited role in long-term protection of NR participants, given the declining serum neutralising capacity and anti-E1E2 antibody levels between clearance and follow-up. These findings align with the work by Nishio et al,⁵² who observed a lack of LLPCs but persisting functional memory B cells after treatment-induced HCV cure. Following antibody waning, antigen exposure-induced recall of memory B cells imprinted during primary infection could be an important mechanism to renew previously generated protective antibody responses. Indeed, such potentially protective recall may have been stimulated in NR participants by undetected transient episodes of spontaneously resolved viraemia during follow-up.

Alternatively, some individuals’ lack of reinfection might stem from behavioural changes within the cohort. Behavioural risk fluctuated substantially over time, hindering attempts to quantify the contribution of lower-risk behaviour to remaining reinfection-free. Acknowledging this fluctuation, Hage et al⁵³ recently classified longitudinal HCV risk behaviour of a subset of our study participants as predominantly lower-risk or predominantly higher-risk. Importantly, all participants who became reinfected despite predominantly lower-risk behaviour were assigned to LP1/LP2, that is, profiles characterised by weaker responses, whereas 2/5 participants who remained non-reinfected despite predominantly higher-risk behaviour were assigned to LP4, that is, the strongest anti-HCV profile. Thus, the identified immunological profiles appear to provide explanatory information about reinfection occurrence beyond the insights available from monitoring only behaviour.

To summarise, we identified key components of an antibody profile associated with protection against observed HCV reinfection. Our findings demonstrate that antibody responses at the time of primary infection clearance may be important for conferring protection against reinfection. These potentially protective responses apparently target multiple epitopes, for example, AR4 and AS412, and may require the induction of effector functions such as ADCP, ADCC and ADCD. This study provides valuable insights to inform the design of effective HCV vaccines.

Data availability statement

Data are available upon reasonable request.