Development and Validation of Novel Models Based on Clonality Immunoglobulin Gene Rearrangement for Evaluation of Bone Marrow Involvement and Prognostic Prediction in Patients with Diffuse Large B-cell Lymphoma: a Multicenter Retrospective Study

Ze-Lin Liu ¹ Qianping Li ² Rui-Qi Liu ³ Yang-Yang Ding ¹ Ya Liao ¹ Lin-Hui Hu ¹ Lian-fang Pu ¹ Chun-Hua Zhu ⁴ Sheng-Hui Zhang ^2* Shudao Xiong ^1*

• ¹ Department of Hematology, The Second Hospital of Anhui Medical University, Hefei, China
• ² Wenzhou Medical University, Wenzhou, Zhejiang Province, China
• ³ Qilu Hospital, Shandong University, Jinan, Shandong Province, China
• ⁴ Air Force HealthCare Center for Special Services Hangzhou, Hangzhou, Jiangsu Province, China

Bone marrow involvement (BMI) is a poor prognostic factor in diffuse large B cell lymphoma (DLBCL), and accurate evaluation of BMI is crucial for determining stages and prognosis. This study aimed to identify the most effective examinations for evaluating BMI in DLBCL, including positron emission tomography-computed tomography (PET/CT), immunoglobulin gene rearrangement (IGR), flow cytometry (FCM), bone marrow cytology (BMC) and bone marrow biopsy pathology (BMB), and to further explore its prognostic significance in DLBCL patients. A total of 364 newly diagnosed patients with DLBCL were included in this retrospective study.Compared to other detection methods, Clonal IGR BMI-positive were found the highest rate of 114 patients (31.3%), and IGR BM involvement-positive patients of DLBCL had the worst survival outcomes, especially among patients in stages I to III (P<0.001). Notably, PET/CT existed some limitations in BMI diagnosis, particularly in stage IV patients (P>0.05). Additionally, the combination of IGR and BMB demonstrated superior prognostic predictive capability for the patients in stage IV (P<0.001). Multivariate analysis further confirmed that double-positive BMI of IGR and BMB was an independent prognostic factors of PFS (P=0.026) and OS (P=0.042). In addition, the novel IPI and NCCN-IPI stratification models were established by incorporating the combination of IGR and BMB in training group. The C-index of novel models were increased when IGR and BMB were supplemented in our cohort. Our results suggest that IGR is the most valuable methods for evaluating BMI compared to traditional detection methods. Adding the combination of IGR and BMB to the IPI and NCCN-IPI score may improve their predictive ability.In summary, IGR is ensential for evaluation of BMI, and provide an ideal method for disease staging and risk stratification in DLBCL patients in the rituximab era.