A biomarker analysis of tumor samples from two landmark first-line extensive-stage small-cell lung cancer (ES-SCLC) trials, IMpower133 and SKYSCRAPER-02, reveals that B7-H3 is highly and broadly expressed across SCLC molecular subtypes, but does not function as a prognostic marker for survival outcomes. These findings have implications for the development of B7-H3–targeted therapies in SCLC.

Clinical Takeaway

B7-H3 is expressed in >80% of pretreatment ES-SCLC tumor samples and is uniformly distributed across all four SCLC molecular subtypes, making it a broadly accessible therapeutic target. However, high versus low B7-H3 expression does not predict improved progression-free survival (PFS) or overall survival (OS) in patients treated with either carboplatin/etoposide (CE) alone or in combination with atezolizumab, suggesting B7-H3 is a targetable antigen rather than a prognostic biomarker in this setting.

Background: B7-H3 in SCLC

B7-H3 (also known as CD276) is a cell surface protein belonging to the B7 immune checkpoint family. It is frequently overexpressed in multiple solid tumor types, including lung cancers, and has been implicated in tumor progression and immune evasion. Interest in B7-H3 as a therapeutic target has grown substantially with the development of B7-H3–directed antibody-drug conjugates and bispecific antibodies.

Study Design

  • Study type: Biomarker analysis using tumor samples from two prospective, randomized trials: IMpower133 (NCT02763579) and SKYSCRAPER-02 (NCT04256421)
  • Biomarker-evaluable population: 462 pretreatment tumor samples underwent bulk RNA sequencing (RNA-seq)
  • Treatment arms analyzed: Patients receiving carboplatin/etoposide (CE) or atezolizumab plus CE (ACE)
  • B7-H3 protein expression: Assessed by immunohistochemistry (IHC) on 5-micron paraffin sections using heat-induced retrieval and peroxidase-conjugated secondary detection
  • Outcome dichotomization: Samples were categorized as high or low B7-H3 expression based on median gene expression level
  • Combined dataset: Generated using batch correction to account for sample processing differences between studies
  • Molecular subtyping: Performed as previously described by Nabet et al. (Cancer Cell, 2024), with correlations investigated between B7-H3 and transcription factors, cell surface targets, and SCLC molecular subtypes

B7-H3 Expression Findings

  • Prevalence: B7-H3 protein expression by IHC was detected in >80% of pretreatment ES-SCLC tumor samples
  • Molecular subtype distribution: RNA expression was consistent across all four SCLC molecular subtypes, indicating broad expression regardless of tumor subtype
  • Transcription factor correlations: At the RNA level, B7-H3 expression showed no correlation with SCLC-defining transcription factors or canonical cell surface targets
  • PD-L1 correlation: RNA-seq data demonstrated no correlation between B7-H3 expression and PD-L1 expression as assessed by IHC, suggesting independent expression dynamics

Prognostic Outcomes by B7-H3 Expression

  • IMpower133 — Atezolizumab + CE arm (n=132; OS data cut: January 24, 2019):
    • Median PFS: 5.5 months (high B7-H3) vs 4.4 months (low B7-H3); HR 0.76 (95% CI: 0.53–1.09)
    • Median OS: 12.6 months (high B7-H3) vs 9.6 months (low B7-H3); HR 0.72 (95% CI: 0.48–1.06)
  • SKYSCRAPER-02 — Atezolizumab + CE arm (n=110; OS data cut: February 6, 2022):
    • Median PFS: 5.7 months (high B7-H3) vs 4.4 months (low B7-H3); HR 0.98 (95% CI: 0.65–1.49)
    • Median OS: 12.9 months (high B7-H3) vs 11.5 months (low B7-H3); HR 0.79 (95% CI: 0.51–1.22)
  • IMpower133 — CE alone arm (n=139):
    • Median PFS: 4.4 months (high B7-H3) vs 4.3 months (low B7-H3); HR 1.0 (95% CI: 0.71–1.41)
    • Median OS: 9.7 months (high B7-H3) vs 10.1 months (low B7-H3); HR 1.12 (95% CI: 0.77–1.62)

Across all three treatment cohorts, confidence intervals crossed 1.0 for both PFS and OS hazard ratios, confirming the absence of statistically significant prognostic impact for B7-H3 expression level.

Key Clinical Implications

✔ B7-H3 is broadly expressed in ES-SCLC, with IHC positivity exceeding 80% of pretreatment samples, suggesting it may serve as a widely accessible target for directed therapies across the ES-SCLC population.

✔ Uniform B7-H3 RNA expression across all four SCLC molecular subtypes indicates that subtype-based patient selection may not be necessary for B7-H3–targeted treatment strategies.

✔ The lack of correlation between B7-H3 and PD-L1 expression suggests that B7-H3–directed therapies may offer a complementary or independent mechanism relative to existing immune checkpoint inhibition in SCLC.

✔ B7-H3 expression level (high vs. low) is not a prognostic factor for PFS or OS in ES-SCLC patients receiving either chemotherapy alone or chemotherapy plus atezolizumab, meaning it should not be used to identify patients with inherently better or worse prognosis.

✔ Future clinical development of B7-H3–targeted agents in SCLC should focus on B7-H3 as a tumor-associated antigen for therapeutic targeting rather than as a biomarker for prognostic stratification or patient selection based on expression level.

Bottom Line

This biomarker analysis from the IMpower133 and SKYSCRAPER-02 trials establishes that B7-H3 is highly and uniformly expressed in ES-SCLC, irrespective of molecular subtype, transcription factor profile, or PD-L1 status. Elevated B7-H3 expression does not confer a survival advantage or disadvantage, confirming its status as a non-prognostic, broadly expressed tumor antigen. These data support the rationale for pursuing B7-H3–directed therapeutic strategies such as antibody-drug conjugates or T-cell–engaging bispecifics across an unselected ES-SCLC population, without the need for expression-level–based patient stratification.

Sources:

  • Presented as a biomarker analysis from the IMpower133 (NCT02763579) and SKYSCRAPER-02 (NCT04256421) clinical trials. Expression and Prognostic Impact of B7-H3 in Patients With Extensive-Stage Small-Cell Lung Cancer. Data referenced from final OS data cuts: IMpower133 (January 24, 2019) and SKYSCRAPER-02 (February 6, 2022).
  • Nabet, et al. (2024). SCLC molecular subtype classification. Cancer Cell.