免疫组化在 DLBCL 中进行细胞起源分类的应用、挑战与展望：一项针对美国肿瘤科医生和病理科医生的调查

Background: Cell of Origin (COO) classification in treatment-naïve (TN) diffuse large B cell lymphoma (DLBCL) is universally recommended across guidelines as an essential prognostic tool, yet some patients are never tested or receive equivocal results that prohibit classification. We sought to determine what barriers to COO classification exist and what could mitigate these barriers. Methods: Double-blinded phone interviews with 8 oncologists and 10 pathologists informed current practices, challenges, and survey design. A double-blinded, online survey of 60 oncologists who treat DLBCL and 60 pathologists (26 board-certified in hematopathology) who perform COO classification was conducted from September 29 to October 10, 2025. Respondents were from a mix of academic and community-based settings in the United States. Results: We found that oncologists view COO as a valuable tool that provides prognostic insight and a more comprehensive diagnosis, with 54% of oncologists incorporating COO classification into treatment decision-making for some patients and an additional 32% of oncologists incorporating COO for all patients. Survey results suggest that 83% of TN DLBCL patients in the US are tested for COO, with 90% of US practices using immunohistochemical (IHC) techniques. Oncologists and pathologists prefer IHC due to its accessibility, fast turnaround time, and relatively low cost. Gene expression profiling (GEP), while offering more granular classification, is less common due to higher costs and taking 4-6 days longer to return results. 89% of US oncologists were confident using IHC results for treatment selection as they consider the IHC-GEP concordance sufficient. Although IHC is routine, 10-15% of samples initially yield equivocal results due to challenges with all three Hans algorithm markers, particularly with CD10 and MUM1, citing variability in staining protocols, ambiguity in scoring weakly positive cells, and difficulty interpreting focal versus diffuse signal. After secondary review, only ~3% of samples remain equivocal and another ~6% are unreportable due to inadequate sample quality or quantity. The potential for oncologists to use pathology results is impeded by missing information in the report, lack of a COO classification in the top-line diagnosis, and variation in report structure, even among pathologists at the same institution. Conclusion: IHC is a widely used, reliable method for COO classification in TN DLBCL with 86% of oncologists using the results for treatment decisions. Our findings suggest that developing consensus guidelines for standardized pathology reports and developing educational materials to reinforce the clinical implications of COO and guide handling of equivocal results will help mitigate testing barriers and increase the proportion of patients receiving actionable COO classifications.