Peginterferon alfa-2b remains one of the few finite treatment options available for selected patients with chronic hepatitis B, particularly in strategies aimed at durable immune control of the virus. Yet the drug carries a well-recognized hematologic price: leukopenia, neutropenia, and thrombocytopenia are among its best-documented adverse effects and can shape treatment tolerability, monitoring intensity, dose decisions, and the need for supportive care. A new exploratory analysis, published in Frontiers in Medicine, has now tracked these blood count declines in detail within a small, prospectively followed cohort of 21 Chinese patients, quantifying how uniformly and how deeply counts fell during the first ten weeks of therapy, and asking whether the pattern differed between patients who showed an early serological response and those who did not.
The study, conducted at the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University between July 2024 and February 2025, enrolled adults with chronic hepatitis B who received peginterferon alfa-2b at 180 micrograms subcutaneously once weekly, with or without background oral nucleos(t)ide analogue therapy. The investigators classified patients operationally by early serological response through Week 15: among those with detectable hepatitis B surface antigen (HBsAg) at baseline, response required either a decline of at least 1.0 log10 IU/mL from baseline or a fall of HBsAg below the assay’s lower limit of detection; among those already below that limit, response required anti-HBs seroconversion to at least 10 mIU/mL. Thirteen patients met response criteria and eight did not. Because the cohort size was set by the available parent study rather than by a formal sample-size calculation, the authors framed the entire analysis as hypothesis-generating.
The central clinical finding was strikingly uniform. White blood cell count, absolute neutrophil count, and platelet count all declined from the first visit to the third visit in both response groups, and every single paired change was negative. Among the 13 responders, the median paired changes from Visit 1 to Visit 3 were −2.430 × 10⁹/L for white blood cells (bootstrap 95% confidence interval: −2.530 to −1.530), −1.730 × 10⁹/L for neutrophils (−1.990 to −1.094), and −67.0 × 10⁹/L for platelets (−83.0 to −55.0), each with a matched-pairs rank-biserial correlation of −1.00. Among the 8 non-responders, the corresponding medians were −2.745 × 10⁹/L for white cells (−3.570 to −2.190), −1.870 × 10⁹/L for neutrophils (−2.260 to −1.110), and −97.5 × 10⁹/L for platelets (−112.0 to −64.0). Visit 2 occurred at a median of 34 days after Visit 1, and Visit 3 at a median of 70 days, corresponding roughly to weeks 5 and 10 of treatment.
The observed nadirs told a similar story of broad, treatment-associated myelosuppression. Median white cell nadirs were 2.29 × 10⁹/L in responders and 3.25 × 10⁹/L in non-responders; median neutrophil nadirs were 1.10 and 1.69 × 10⁹/L respectively; and median platelet nadirs were 86.0 and 112.0 × 10⁹/L. However, when the investigators formally compared the two groups, the estimates were imprecise. The responder-minus-non-responder Hodges–Lehmann shift for white cell nadir was −0.810 × 10⁹/L with a confidence interval spanning −1.240 to 0.160, and for neutrophil nadir −0.391 × 10⁹/L (−0.830 to 0.330); the platelet nadir shift was −20.0 × 10⁹/L (−39.0 to 7.0). None of these comparisons established a response-group-specific toxicity pattern, and the authors were careful to note that the wide intervals preclude any claim of equivalence between the groups as well.
Binary thresholds underscored how common the cytopenias were. At Visit 2, white blood cell counts below 4.0 × 10⁹/L occurred in 19 of 21 patients, or 90.5% (exact 95% confidence interval: 69.6% to 98.8%). Neutrophil counts below 1.0 × 10⁹/L were seen in 7 of 21 patients (33.3%), and platelet counts below 100 × 10⁹/L in 10 of 21 (47.6%). The near-universality of the white cell threshold led the team to treat it as a pragmatic clinical-management context rather than a discriminating molecular endpoint: when almost every patient crosses a cutoff, that cutoff carries little information for distinguishing molecular subgroups within the cohort.
A methodologically notable feature of the study was its handling of supportive care. Oral leukocyte-support medication was given to some patients after their counts fell, which raises the question of whether later blood counts reflect the drug’s toxicity or the mitigation of it. The team reconstructed treatment timing from deidentified records and built a post-documented oral-support-date indicator, classifying each complete blood count as before or after the first documented treatment date. In a random-intercept model covering 60 planned-visit observations from all 21 participants, the post-date term was −0.008 × 10⁹/L for white cells (P = 0.979) and −0.105 × 10⁹/L for neutrophils (P = 0.632), meaning the indicator did not materially alter the fitted decline. Boundary analyses coding ambiguous same-day visits as exposed or unexposed, and a strict pre-support paired set of five responders and four non-responders, all remained directionally consistent with decline. Crucially, because support was initiated in response to low counts, these non-randomized analyses cannot estimate treatment efficacy or remove confounding by indication.
The second half of the study turned to whole-blood RNA sequencing, asking whether baseline host molecular state was associated with the hematologic endpoints. Blood samples collected in PAXgene tubes at Weeks 0, 5, and 10 were sequenced on an Illumina NovaSeq 6000 platform, and gene-level counts were normalized using the variance-stabilizing transformation in DESeq2. Hallmark pathway activity was summarized with gene set variation analysis across 50 pathways per endpoint family, with Benjamini–Hochberg false-discovery-rate correction applied within complete families. The result was a disciplined null: across 350 baseline associations in seven endpoint-specific families, no Hallmark pathway reached the prespecified exploratory threshold of q < 0.10, with a minimum q of 0.251. No baseline candidate-gene association met the threshold either (minimum q = 0.417). In paired Week-10-minus-Week-0 analyses among the 20 patients with both time points, none of 150 Hallmark associations reached q < 0.10.
Because whole-blood expression reflects both cell-intrinsic transcription and shifts in circulating cell composition, the researchers ran a sensitivity analysis using xCell2, a reference-based tool that infers cell-type enrichment from bulk transcriptomes. The inferred neutrophil enrichment score showed only modest and imprecise concordance with measured neutrophil counts at baseline (Spearman rho = 0.382, confidence interval −0.110 to 0.732), but tracked measured counts much more strongly across matched visits (rho = 0.693 for neutrophils, 0.559 for white cells) and for within-person changes over time. Composition adjustment changed several pathway estimates, and a few adjusted platelet-endpoint signals nominally crossed q < 0.10, but none reached q < 0.05, four of seven had nonconverged confidence intervals, and the intervals were extremely wide. The authors declined to interpret these unstable signals as pathway discoveries, emphasizing that the enrichment scores are relative measures derived from the same bulk data as the pathway scores, leaving residual confounding and mathematical coupling as live concerns.
The discussion places the findings in biological context without overclaiming mechanism. Type-I-interferon receptors and signaling have been demonstrated in CD34-positive hematopoietic progenitor cells, and interferon exposure can rapidly suppress hematopoiesis, supporting a marrow-production component for the leukocyte and neutrophil declines. For platelets, experimental type-I-interferon signaling can reduce platelet production and function, and prior small studies in chronic hepatitis B and other viral hepatitides have suggested impaired platelet production rather than antibody-mediated destruction. The authors note, however, that they did not measure bone-marrow function, endogenous G-CSF, antiplatelet antibodies, or thrombopoiesis in their own cohort, so these mechanisms remain contextual rather than demonstrated.
The study’s limitations are substantial and openly stated. It is an exploratory secondary analysis of a single-center cohort of 21 participants with no independent validation; the between-group confidence intervals are wide; only one participant was HBeAg-positive, limiting generalization; and whole-blood transcriptomics cannot be read as liver-intrinsic biology. What the analysis delivers is a carefully quantified clinical picture: during peginterferon alfa-2b-based therapy for chronic hepatitis B, white cell, neutrophil, and platelet counts fell in every patient studied, the decline was not clearly modified by early serological response status, and a near-universal white cell threshold is more useful for clinical management than for molecular stratification. The transcriptomic null results, reported with complete multiplicity control, serve as a cautionary benchmark for future biomarker claims in this space, and the authors stress that all findings require confirmation in larger, independent, preferably multicenter cohorts with orthogonal measurements of cell composition and hematopoietic function.
Subject of Research: Hematologic safety and whole-blood transcriptomics during peginterferon alfa-2b therapy in chronic hepatitis B
Article Title: Longitudinal hematologic decline during peginterferon alfa-2b therapy in chronic hepatitis B: clinical safety and exploratory whole-blood transcriptomics
Article References: Li, J., Chen, W., Chen, J., Xing, J., Chen, Q., Zhang, J., Shi, R., Pan, C., & Chen, Y. (2026). Longitudinal hematologic decline during peginterferon alfa-2b therapy in chronic hepatitis B: clinical safety and exploratory whole-blood transcriptomics. Frontiers in Medicine, 13, Article 1919392. https://doi.org/10.3389/fmed.2026.1919392
Image Credits: AI Generated
DOI: 10.3389/fmed.2026.1919392
Keywords: chronic hepatitis B, peginterferon alfa-2b, leukopenia, neutropenia, thrombocytopenia, hematologic safety, whole-blood RNA-seq, transcriptomics, HBsAg response, cell composition, exploratory cohort study, Longitudinal
News Source: Kristina Jarvis. (October 26, 2026). Blood Counts Fall Uniformly During Peginterferon Therapy for Chronic Hepatitis B, Small Study Finds. Scienmag.
