Background: Rare hematologic diseases are frequently underdiagnosed or misdiagnosed due to their clinical complexity. Whether new-generation large language models (LLMs), particularly those using chain-of-thought reasoning, can improve diagnostic accuracy remains unclear. Objective: This study aimed to evaluate the diagnostic performance of new-generation commercial LLMs in rare hematologic diseases and to determine whether the LLM output enhances physicians’ diagnostic accuracy. Methods: We conducted a 2-phase study. In the retrospective phase, we evaluated 7 mainstream LLMs on 158 nonpublic real-world admission records covering 9 rare hematologic diseases, assessed diagnostic performance using top-10 accuracy and mean reciprocal rank (MRR), and evaluated ranking stability via Jaccard similarity and entropy. Spearman rank correlation was used to examine the association between physicians’ diagnoses and LLM-generated outputs. In the prospective phase, 28 physicians with varying levels of experience diagnosed 5 cases each, gaining access to LLM-generated diagnoses across 3 sequential steps to assess whether LLMs can improve diagnostic accuracy. Results: In the retrospective phase, ChatGPT-o1-preview demonstrated the highest top-10 accuracy (70.3%) and MRR (0.577), and DeepSeek-R1 ranked second. Diagnostic performance was low for amyloid light-chain (AL) amyloidosis; Castleman disease; Erdheim-Chester disease; and polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes (POEMS) syndrome. Interestingly, higher accuracy often correlated with lower ranking stability across most LLMs. The physician performance showed a strong correlation with both top-10 accuracy (ρ=0.565) and MRR (ρ=0.650). In the prospective phase, LLMs significantly improved the diagnostic accuracy of less-experienced physicians; no significant benefit was observed for specialists. However, when LLMs generated biased responses, physician performance often failed to improve or even declined. Conclusions: Without fine-tuning, new-generation commercial LLMs, particularly those with chain-of-thought reasoning, can identify diagnoses of rare hematologic diseases with high accuracy and significantly enhance the diagnostic performance of less-experienced physicians. Nevertheless, biased LLM outputs may mislead clinicians, highlighting the need for critical appraisal and cautious clinical integration with appropriate safeguard systems.

Background: Previous studies mainly focused on renal impairment (RI) at multiple myeloma (MM) diagnosis, with few investigating new‐onset RI post‐MM diagnosis. This study aims to indicate the incidence, clinical characteristics and predictive value of new‐onset RI in MM patients. Methods: We conducted a multicenter, retrospective cohort study including 1953 newly diagnosed MM patients from West China Hospital from July 1, 2008, to February 30, 2024 and the international MMRF‐CoMMpass database. Among them, 1770 patients received novel therapeutic agents including immunomodulatory drugs (IMiDs) and proteasome inhibitors (PIs) as first‐line therapy. RI was defined as serum creatinine higher than 2 mg/dL or impaired creatinine clearance (< 40 mL/min/1.73m ² ). The association between new‐onset RI and mortality risk was investigated by Kaplan–Meier analysis and Cox proportional hazard models. Results: Of the cohort, 16.6% developed new‐onset RI, with the majority (67.1%) occurring within 2 years post‐MM diagnosis. The median overall survival (OS) was significantly shorter in new‐onset RI than in those without RI (68 vs. 122 months, p < 0.001). New‐onset RI was an independent risk factor for mortality (HR 1.55, 95% CI 1.28–1.88, p < 0.001), and earlier onset of RI was associated with a higher mortality risk. Moreover, patients with recovery of renal function had prolonged OS compared to those without recovery (95 vs. 64.8 months, p = 0.01). Older age, higher stage of the international stage system (ISS) and RI at diagnosis seemed to be risk factors for new‐onset RI, while first‐line therapy with PIs and IMiDs combinations was associated with a lower risk of RI development (HR 0.69, 95% CI 0.51–0.94, p = 0.017). Conclusions: In conclusion, the incidence of new‐onset RI is high in MM, and is a significant risk factor for mortality, posing a substantial threat to MM patients. Early identification of high‐risk patients for new‐onset RI and prompt preventive strategies is critical for improving MM prognosis. Trial Registration: This study was registered by the Chinese Clinical Trial Registry (ChiCTR2400081476, https://www.chictr.org.cn )

Background: Renal impairment is one of the common characteristics of multiple myeloma (MM) and makes management of MM more complicated. Even though monoclonal antibodies targeting CD38 have wildly succeeded in treating MM, the addition of anti-CD38 monoclonal antibodies to standard therapy to treat MM patients with renal insufficiency is still poorly studied. Objectives: This study aims to evaluate whether using anti-CD38 monoclonal antibody-based immunotherapy would improve the prognosis of MM patients with renal insufficiency. Design: This is a systematic review and meta-analysis. Data sources and methods: We searched Scopus, PubMed, Cochrane Central Register of Controlled Trials (CENTRAL), Ovid MEDLINE, Ovid Embase, and Web of Science Core Collection for randomized controlled trials that enrolled patients with MM who received CD38-targeting monoclonal antibody regimens and reported the efficacy and survival of MM with renal insufficiency. We then performed a meta-analysis to estimate the efficacy of adding anti-CD38 monoclonal antibodies to backbone regimens in MM with renal insufficiency. Results: In 7594 studies screened, 12 phase III trials were eligible, including 5 trials for newly diagnosed MM (NDMM; 3194 patients; 1261 with renal insufficiency) and 7 trials for relapsed refractory MM (RRMM; 2657 patients; 648 with renal insufficiency). Among NDMM patients with renal insufficiency, the addition of anti-CD38 monoclonal antibody to backbone regimens was associated with improved progression-free survival (PFS; pooled HR, 0.50; 95% CI, 0.38–0.67; p < 0.001), with little evidence of heterogeneity (Cochran Q, p = 0.19; I² = 34.7%). Similar results were seen among RRMM patients with renal insufficiency (pooled HR, 0.46; 95% CI, 0.37–0.57; p < 0.001), with no evidence of heterogeneity (Cochran Q, p = 0.89; I² = 0%). Similarly, the addition of anti-CD38 monoclonal antibody in RRMM among patients with renal insufficiency was associated with improved overall survival (OS; pooled HR, 0.70; 95% CI, 0.57–0.88; p = 0.002), with no significant heterogeneity (Cochran Q, p = 0.69; I² = 0%). Conclusion: This meta-analysis suggests that the addition of anti-CD38 monoclonal antibodies benefits PFS in both NDMM and RRMM with renal insufficiency and OS in RRMM patients with renal insufficiency.

Disclosures. No relevant conflicts of interest to declare.Figure 1

Background: Multiple myeloma (MM) exhibits significant heterogeneity. Traditional risk stratification often relies on static baseline data, failing to capture the dynamic evolution of biomarkers. This study aimed to develop a prognostic model incorporating longitudinal biomarker trajectories to improve risk stratification for newly diagnosed MM patients. Methods: We retrospectively analyzed 635 patients from West China Hospital between February 1, 2017 and December 30, 2022. Biomarkers including M protein, free light chains (FLC), creatinine (Cr), hemoglobin (Hb), immunoglobulin levels, and others were collected monthly within the first‐year post‐treatment. Patients were randomly divided into an internal training (70%) and validation (30%) cohort. Group‐based trajectory modeling (GBTM) and Kaplan–Meier analysis were used to identify distinct biomarker evolution patterns, and the GBTM risk model was validated by the lowest mean squared error (LMSE) method. Cox regression analysis and a nomogram were employed to develop a prognostic model to provide individualized survival probability estimates and were validated in the internal validation cohort using time‐dependent ROC analysis. Results: GBTM identified dFLC, IgM, Cr, and Hb trajectories as significant prognostic factors. Patients scoring ≥ 3 (high‐risk) showed significantly poorer overall survival than those scoring < 3 (low‐risk) in both the training (median OS 32.2 months vs. not reached, p < 0.0001) and validation cohorts (36.8 months vs. not reached, p < 0.005). Then we established a nomogram that included the GBTM risk group, ISS stage, age, frontline treatment, pulmonary disease, cardiovascular disease, and thrombotic events. The nomogram demonstrated good predictive performance with AUC values consistently above 0.7 for 2‐ and 3‐year survival in both cohorts (Training: 0.71, 0.70; Validation: 0.70, 0.70). Conclusion: Our study successfully developed a prognostic risk stratification model for MM patients using biomarker trajectories. This model may help identify high‐risk patients and predict long‐term survival, offering a valuable tool for personalized treatment planning.