Blood-based scores and immune biomarkers reshape outcome prediction in blood cancers

Immunotherapy has transformed the treatment of blood cancers, but it has also exposed an uncomfortable truth: the same therapy that produces durable remissions in one patient can fail completely, or trigger life-threatening toxicity, in another. A new narrative review published in Biomarker Research on 31 October 2026 brings together the current evidence on prognostic scores and biomarkers in hematologic malignancies, arguing that accurate identification of predictive and prognostic markers is now essential to optimize patient selection, improve therapeutic efficacy, and minimize treatment-related toxicity. The review, led by Régis Costello and colleagues spanning institutions in France, examines immune checkpoint inhibitors, antibody-based therapies, and adoptive cellular immunotherapies built on genetically engineered T cells and natural killer cells, all of which have achieved unprecedented clinical outcomes in several lymphoid and myeloid neoplasms.

The central problem the authors confront is heterogeneity. While some patients experience long-term disease control or even cure, others exhibit primary resistance, fail to achieve meaningful responses, or rapidly relapse following an initial response. Immune-based therapies can also cause potentially life-threatening toxicities, including cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), prolonged cytopenia, infections, and immune-related adverse events. This considerable interindividual variability, the review argues, highlights the urgent need for reliable biomarkers capable of predicting treatment efficacy and toxicity before therapy initiation. In contrast to solid tumors, where biomarkers such as PD-L1 expression, microsatellite instability, and tumor mutational burden have been extensively investigated, predictive biomarkers in hematologic malignancies remain less clearly established. The biological heterogeneity of blood cancers, coupled with their unique interactions with the immune system, presents additional challenges for biomarker development and validation, and few markers have demonstrated consistent predictive value across different immunotherapeutic modalities.

The review’s structured literature search covered publications from January 2010 to September 2026, with particular emphasis on studies published between January 2021 and September 2026, using PubMed/MEDLINE as the primary database and supplementing with citation tracking. Because the manuscript was designed as a narrative review rather than a systematic review or meta-analysis, the authors acknowledge the search was intended to provide broad and clinically relevant coverage of an evolving field rather than to constitute an exhaustive PRISMA-compliant systematic search. The authors distinguish between established prognostic associations, emerging biomarkers, and biomarkers with demonstrated predictive potential, a distinction they stress is critical: a prognostic biomarker provides information about clinical outcome irrespective of the treatment received, whereas a predictive biomarker identifies patients more or less likely to benefit from a specific therapeutic intervention. Several parameters described in the review currently have predominantly prognostic evidence; an association with progression-free or overall survival does not by itself demonstrate treatment-specific predictive value.

Classical clinical prognostic scores remain central to the review’s assessment. The International Prognostic Index (IPI), developed in 1993 for aggressive B-cell lymphomas, incorporates five adverse factors: age over 60 years, elevated serum lactate dehydrogenase, ECOG performance status of 2 or higher, Ann Arbor stage III-IV, and more than one extranodal disease site. In the CAR-T cell therapy era, several studies have shown that a high IPI remains associated with inferior progression-free survival and overall survival, with analyses of patients receiving anti-CD19 CAR-T cells demonstrating that elevated IPI scores correlate with increased tumor burden, reduced CAR-T expansion, and higher relapse rates. The R-IPI, developed after rituximab entered lymphoma treatment, has shown utility in identifying diffuse large B-cell lymphoma patients who derive durable benefit from CAR-T therapy, though real-world studies report mixed and sometimes contradictory findings regarding its prognostic value in this setting. The NCCN-IPI, which refines risk stratification with detailed age categories and LDH stratification, has shown superior prognostic performance compared with the traditional IPI among patients receiving contemporary cellular therapies.

For follicular lymphoma, the FLIPI score, designed for chemoimmunotherapy-treated patients, does not predict outcomes in those receiving modern immunotherapeutic approaches such as CAR-T cells, although patients with high FLIPI scores frequently exhibit increased disease burden and a more immunosuppressive tumor microenvironment. In chronic lymphocytic leukemia, the CLL-IPI combines age, clinical stage, β2-microglobulin, TP53 status, and IGHV mutational status; emerging studies of CAR-T therapies and bispecific antibodies suggest high-risk patients demonstrate lower response durability and shorter survival, likely reflecting adverse biological features rather than immunotherapy resistance alone, though the review notes this score is not the most relevant for prognosis compared with memory-like T-cell counts, transcriptional studies, and exhausted phenotypes. In multiple myeloma, the International Staging System remains predictive in patients receiving BCMA-targeted therapies, with ISS stage III patients generally experiencing higher tumor burden, lower response durability, and inferior progression-free survival following CAR-T therapy, though the authors note the ISS does not really seem to be a strong predictor of response. The R-ISS and R2-ISS, which incorporate high-risk cytogenetics and LDH, were developed in patients treated with conventional therapies and have not been formally validated as CAR-T-specific prognostic models.

Peripheral blood-derived inflammatory scores receive particular attention because they are inexpensive, reproducible, and readily available in routine clinical practice. These include the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), lymphocyte-to-monocyte ratio (LMR), systemic immune-inflammation index (SII), prognostic nutritional index (PNI), and Glasgow Prognostic Score. The biological rationale lies in the balance between protumoral inflammatory processes and antitumor immune surveillance: elevated neutrophils, monocytes, and platelets may promote tumor growth through cytokine secretion, angiogenesis, and suppression of cytotoxic lymphocyte activity, whereas lymphocytes represent the cornerstone of adaptive antitumor immunity. In classical Hodgkin lymphoma, large multicenter studies demonstrated that an NLR above 6 independently predicts inferior progression-free and overall survival, yet evidence supporting NLR as a predictive biomarker of response to nivolumab or pembrolizumab remains limited. In DLBCL, high NLR predicts adverse outcomes following standard immunochemotherapy with R-CHOP and may also influence responses to CAR-T-cell therapy, with several studies suggesting elevated pre-infusion NLR is associated with inferior expansion and persistence of CAR-T cells.

The SII, which combines platelets, neutrophils, and lymphocytes into a single parameter, has demonstrated prognostic value across multiple hematologic malignancies. In DLBCL, elevated SII predicts inferior survival independently of IPI and molecular subtype, a finding confirmed by a meta-analysis of five articles including 592 cases. In non-M3 acute myeloid leukemia, a retrospective cohort found that a high baseline SII was significantly associated with inferior overall and event-free survival and remained an independent prognostic factor after adjustment for established clinical and laboratory variables, though the authors caution that current evidence is limited to a small number of retrospective studies requiring validation in large prospective multicenter cohorts. The Pan-Immune-Inflammation Value integrates neutrophil, platelet, monocyte, and lymphocyte counts, and the PILE score further adds lactate dehydrogenase and ECOG performance status; studies in primary central nervous system lymphoma and diffuse large B-cell lymphoma have reported associations with shorter overall and progression-free survival, with PILE also associated with treatment response in primary CNS lymphoma.

Endothelial activation scores represent another frontier. EASIX, calculated as LDH times creatinine divided by platelet count, serves as an indirect surrogate of systemic endotheliopathy using routinely measured laboratory parameters. Following CAR-T-cell activation, inflammatory mediators including IL-1β, IL-6, IFN-γ, and TNF-α activate and injure vascular endothelial cells, resulting in increased vascular permeability, disruption of endothelial junctions, platelet consumption, and microvascular dysfunction. In the central nervous system, endothelial activation and pericyte dysfunction compromise blood-brain barrier integrity, promoting leakage of plasma proteins, cytokines, and immune cells into the CNS and contributing to cerebral edema and ICANS. In the study of Galli and colleagues, baseline mEASIX, which replaces creatinine with CRP, was predictive for consumptive coagulopathy, CRS graded 2 or higher, and progression-free and overall survival. Pennisi and colleagues found that elevated pre-lymphodepletion EASIX was significantly associated with grade 3 or higher CRS and inferior survival outcomes in patients receiving CD19-directed CAR-T therapy, with mEASIX predicting severe CRS and ICANS with performance comparable to more sophisticated cytokine panels. Limitations persist: most studies are retrospective with relatively small cohorts, cut-off values vary considerably across publications, and EASIX may be influenced by factors unrelated to endothelial dysfunction, including renal impairment, hemolysis, infection, and advanced tumor burden.

The CAR-HEMATOTOX score, specifically designed to predict severe hematologic toxicity following CAR-T-cell therapy, combines five routinely available laboratory parameters measured before lymphodepletion: absolute neutrophil count, hemoglobin, platelet count, CRP, and ferritin. The rationale rests on the observation that prolonged cytopenias following CAR-T therapies are not solely attributable to lymphodepleting chemotherapy but are also influenced by pre-existing bone marrow reserve, systemic inflammation, and immune dysregulation before infusion. Patients classified as high-risk exhibit substantially increased risks of prolonged neutropenia, severe thrombocytopenia, severe anemia, febrile neutropenia, infectious complications, and intensive care admission. In CD19 CAR-T therapy for acute lymphoblastic leukemia, low scores were associated with lower rates of delayed neutrophil recovery, fewer severe infections, higher response rates, higher overall and event-free survival, and lower non-relapse mortality, with survival differences remaining significant after multivariable adjustment. In multiple myeloma BCMA-directed CAR-T therapy, high-risk patients displayed prolonged severe neutropenia, increased severe infection rates, more severe ICANS, and higher one-year non-relapse mortality predominantly attributable to fatal infections, along with inferior progression-free and overall survival.

Beyond these scores, the review surveys emerging biomarker models built on T-cell fitness, NK-cell competency, and cytokine profiling. Baseline characteristics of circulating T cells strongly influence outcomes after CAR-T therapy, bispecific antibodies, or checkpoint blockade, with composite scores incorporating favorable features such as high proportions of stem-cell memory T cells and preserved CD27 and CD28 expression, and unfavorable features including high PD-1, TIGIT, LAG-3, and TIM-3 expression. An immune age score, calculated as senescent plus exhausted T cells divided by naive plus recent thymic emigrants, was associated with progression-free and overall survival in myeloma even after adjusting for performance status, ISS stage, and frailty status. Cytokine signatures, including post-perfusion elevations of IL-6, IL-10, IFN-γ, MCP-1, GM-CSF, and TNF-α, have been associated with severe CRS, severe ICANS, and treatment failure, though these signatures have not been formally standardized as defined scores. The authors conclude that widespread implementation remains constrained by substantial heterogeneity among studies, and that future efforts should focus on prospective multicenter validation, harmonization of scoring systems, and dynamic models integrating conventional laboratory parameters with high-dimensional immune profiling, transcriptomic, spatial, and functional approaches.

Subject of Research: Prognostic scores and biomarkers for predicting immunotherapy outcomes and toxicity in hematologic malignancies

Article Title: Prognostic scores and biomarkers in hematologic malignancies: current evidence and future perspectives

Article References: McDonald, E., Marcenaro, E., Robert, P., Rihet, P., Théodoly, O., & Costello, R. (2026). Prognostic scores and biomarkers in hematologic malignancies: current evidence and future perspectives. Biomarker Research, 14(1), Article 157. https://doi.org/10.1186/s40364-026-01033-8

Image Credits: AI Generated

DOI: 10.1186/s40364-026-01033-8

Keywords: hematologic malignancies, immunotherapy, biomarkers, CAR-T cell therapy, prognostic scores, NLR, EASIX, CAR-HEMATOTOX, cytokine release syndrome, DLBCL, multiple myeloma, immune fitness