PREreview of Stem-like CD8+ T cells preserve HBV-specific responses in HBV/HIV co-infection

PAPER SUMMARY

In their research article, Preechanukul et. al examine CD8+ T cell responses in chronic HBV infection with and without HIV co-infection. Chronic HBV affects approximately 10% of people living with HIV, yet these co-infected individuals are frequently excluded from functional cure studies. This study investigated whether, contrary to conventional expectations, long-term suppressive therapy might preserve rather than impair HBV-specific immune responses in co-infection. The researchers analyzed CD8+ T cell profiles in 61 participants (HBV n=20, HBV/HIV n=20, HIV n=21) using transcriptomic, proteomic, and functional analyses. They assessed exhaustion markers and virus-specific capabilities in individuals on suppressive antiviral therapy. Their findings revealed distinct transcriptomic signatures in co-infected individuals, with upregulation of genes associated with "precursor exhausted" T cells that maintain plasticity despite chronic antigen exposure. These results suggest that individuals with well-controlled HBV/HIV co-infection maintain more functional CD8+ T cell responses. The study suggests approaches for immune restoration in chronic viral infections. However, several methodological limitations require additional clarification. If these recommendations are implemented, it will greatly improve the manuscript and the impact of the work.

MAJOR REVISIONS

Infection Route and Timing Differences

Critique: Further examination is warranted regarding the fundamental difference in infection routes and timing between the study groups, which represents a significant alternative explanation for the observed immunological differences. As the authors briefly acknowledge, HBV mono-infection often results from vertical transmission in early childhood, while HBV/HIV co-infection typically occurs through sexual transmission in adulthood. This distinction is critical because vertically-acquired HBV often leads to immune tolerance mechanisms that are not present in adult-acquired infection. The contrasting immunological imprinting that occurs during these different acquisition scenarios could independently explain the observed differences in exhaustion profiles and T cell functionality, yet the study design does not systematically address this alternative explanation. The immune responses to adult-acquired HBV infection are typically more robust initially, potentially leading to different exhaustion trajectories even before treatment initiation.

Framework for addressing:

• Stratify analysis by known or inferred infection route (vertical vs. horizontal transmission) where this information is available.

• Conduct a subgroup analysis comparing HBV mono-infected patients with confirmed adult-acquired infection to those with HBV/HIV co-infection to control for the timing of infection.

• Include age at infection as a covariate in multivariate analyses. 

• More prominently acknowledge this fundamental limitation and discuss how it impacts the interpretation of results, particularly regarding the potential role of immune tolerance versus exhaustion.

Control for Treatment Duration Disparity

Critique: A central consideration for this study is the substantial difference in treatment duration between the HBV/HIV co-infection group (average 14 years) and HBV mono-infection group (average 4 years), which represents a significant confounding variable that requires robust statistical control. While the manuscript acknowledges this difference and demonstrates correlations between treatment duration and immunological parameters, the study design makes it challenging to distinguish whether the observed immunological differences are attributable to infection status or simply longer viral suppression. This confound is particularly impactful because the authors' own data show that treatment duration positively correlates with both Tpex frequencies and functional HBV-specific responses, suggesting that the key findings could be driven by treatment duration rather than co-infection status. The authors note that "early intervention likely preserves the Tpex pool before terminal exhaustion occurs," yet this explanation supports treatment duration, not HIV co-infection, as the primary mechanism.

Framework for addressing:

• Conduct a subgroup analysis matching participants from both groups with similar treatment durations to determine if immunological differences persist when controlling for this variable.

• Perform multivariate regression analyses that simultaneously evaluate the effects of both treatment duration and infection status on key outcome measures (Tpex frequencies, functional responses).

• Consider propensity score matching to create statistically comparable groups based on treatment duration and other clinical variables.

• Clearly acknowledge the possibility that treatment duration rather than co-infection status may be the primary driver of observed differences.

• If possible, analyze longitudinal samples from a subset of patients to assess the evolution of exhaustion markers over treatment time.

Single-Cell RNA Sequencing Sample Size Limitations

Critique: The single-cell RNA sequencing analysis is conducted on a very limited subsample (n=5 HBV/HIV co-infection, n=6 HBV mono-infection) without justification for this sample size or power calculations. This represents a significant limitation as transcriptomic analyses typically exhibit high inter-individual variability, requiring larger sample sizes to achieve adequate statistical power. With the current sample size, the study likely has insufficient power to detect differentially expressed genes with moderate effect sizes, substantially increasing the risk of both false positives and false negatives. While single-cell technologies provide high cellular resolution, they do not overcome the fundamental requirement for adequate biological replication. This limitation impacts the reliability of the transcriptomic signatures and pathway enrichment analyses presented, which form a foundation for the study's conclusions about differential CD8+ T cell states in mono-infection versus co-infection.

Framework for addressing:

• Conduct and report formal power analyses to establish the minimum detectable effect size given the current sample size for the transcriptomic analyses.

• Validate key differentially expressed genes using qPCR in the larger cohort of 20 patients per group to confirm that findings from the small subsample generalize to the full study population.

• Apply more stringent statistical thresholds for differential expression analysis to reduce false discovery rates, and clearly report these thresholds.

• Employ bootstrapping or similar resampling approaches to assess the stability of differential expression results.

• Explicitly acknowledge the limited statistical power of the transcriptomic analysis as a major limitation in the discussion section.

Effect Size Reporting and Statistical Interpretation

Critique: The statistical reporting in this manuscript would benefit from comprehensive inclusion of effect sizes and confidence intervals to complement the p-values currently presented. While statistical significance testing (p-values) indicates the probability of observed differences occurring by chance, effect sizes provide essential information about the magnitude and biological relevance of these differences. This additional information is particularly important in studies with modest sample sizes, where statistical significance may be achieved despite relatively small effect sizes that might have limited clinical or biological significance. Throughout the manuscript, the authors report statistically significant differences in cell frequencies, gene expression, and functional responses between groups, but the practical importance of these differences remains difficult to assess without standardized effect size measures and their precision (95% confidence intervals). For correlation analyses only significance indicators are visible without reporting actual correlation coefficients in the text, making it difficult to evaluate the strength of these relationships.

Framework for addressing:

• For all key comparisons between patient groups, report appropriate standardized effect size measures (Cohen's d or Hedges' g) alongside p-values.

• Include 95% confidence intervals for all reported means, proportions, and effect sizes to indicate precision.

• For correlation analyses report exact correlation coefficients (r values) in the text or in a supplementary table.

• Distinguish between statistically significant findings with large, moderate, and small effect sizes in the interpretation.

• Consider including a supplementary table summarizing all statistical tests performed with complete reporting of test statistics, degrees of freedom, effect sizes, and confidence intervals.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they used generative AI to come up with new ideas for their review.