Toward individualized prognosis in primary breast diffuse large B-cell lymphoma: incidence trends and a predictive model

Introduction

Primary breast diffuse large B-cell lymphoma (PB-DLBCL) is an exceptionally rare form of extranodal lymphoma, representing a small fraction of both breast malignancies and non-Hodgkin lymphomas (1-4). Its low incidence has hindered comprehensive epidemiologic characterization and limited our understanding of long-term patient outcomes. As a result, clinicians often rely on extrapolated data from DLBCL occurring at more typical sites or small institutional series, which may not reflect the true survival dynamics in this population.

While traditional survival analyses provide an overall picture of prognosis, they do not capture how the probability of survival changes over time for patients who have already survived a given period (5-7). Conditional survival (CS) analysis addresses this gap by offering dynamic, time-updated estimates of survival, which can inform follow-up strategies, counseling, and therapeutic decision-making in real time (8,9). In parallel, identifying the most meaningful combination of prognostic factors remains a key challenge for rare diseases like PB-DLBCL. Best subset regression (BSR) provides a data-driven approach to select an optimal variable set that balances model simplicity and predictive accuracy (10-12). When incorporated into a nomogram, these variables enable individualized survival prediction and risk stratification, translating statistical findings into clinically actionable tools.

In the present study, we leveraged population-level data from 2000 to 2021 to systematically investigate incidence trends of PB-DLBCL, quantify dynamic CS, and develop a BSR-guided nomogram for individualized risk assessment. The Surveillance, Epidemiology, and End Results (SEER) database was selected for its strengths, including a large, population-based cohort and long-term follow-up, which are particularly valuable for studying rare tumors, while we also acknowledge its limitations, such as lack of detailed chemotherapy regimens and molecular data. By combining epidemiologic insights with personalized prognostic modeling, our study aims to provide clinicians with a practical framework to better understand and manage this rare but aggressive lymphoma. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1976/rc).

Methods

The present analysis was conducted using data from the SEER program of the National Cancer Institute, which encompasses population-based cancer registries and represents nearly 28% of the U.S. population. Patients diagnosed between January 2000 and December 2020 were identified from the SEER registries database. Eligible cases were required to have the breast as the primary site (ICD-O-3 site codes C50.0–C50.9) and histological confirmation of diffuse large B-cell lymphoma, not otherwise specified (DLBCL, NOS; ICD-O-3 code 9680/3). Only patients with the first primary malignancy were included (sequence number = “one primary only” or “first of multiple primaries”), and all diagnoses had to be histologically confirmed. Patients were excluded if they had missing age information, lacked pathological confirmation, or had incomplete essential clinicopathological or survival data. To ensure data accuracy, individuals whose diagnosis was established only at autopsy or through death certificates were also excluded. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

From the database, we collected comprehensive clinical information, including demographic variables, Ann Arbor stage at presentation, treatment modalities such as surgery, radiotherapy or chemotherapy, and survival outcomes. Continuous variables, such as age, were categorized into clinically meaningful groups (<50, 50–59, 60–69, 70–79, and ≥80 years) based on previously published studies for inclusion in the nomogram. Descriptive statistics were applied to summarize patient characteristics. The primary outcome was overall survival (OS), defined as the time from diagnosis to death from any cause or last follow-up. Patients who were alive at the last follow-up were censored.

Statistical analysis

All incidence rates were age-adjusted to the 2000 U.S. standard population. Age-standardized rates were derived from the SEER Research Data (Incidence, 17 Registries, November 2024 Submission) and analyzed using SEER*Stat software. Results were reported as cases per 100,000 individuals. The analytic cohort used for survival, CS, and nomogram analyses is a subset of SEER cases filtered by histologic confirmation, complete variables, and exclusion of autopsy/death-certificate-only cases.

Subsequently, we further characterized the CS patterns of this patient population. CS provides dynamic estimates of prognosis by calculating the probability of surviving an additional number of years, given that a patient has already survived a certain period of time after diagnosis. The 10-year CS probabilities were derived using the standard CS formula: CS(y∣x) = S(y+x)/S(x) , where S(x) represents the actuarial survival probability at time x, and CS(y∣x) denotes the likelihood of surviving an additional y years, conditional on having already survived x years (9).

To enable individualized prognostic evaluation, we developed a CS nomogram model that integrated dynamic CS probabilities with a nomogram framework. The study population was randomly divided into training and validation cohorts in a 7:3 ratio. In the training set, BSR was initially applied to identify the most informative prognostic variables, followed by stepwise backward Cox regression to further refine the selection. We also performed multivariable Cox regression analysis to evaluate the prognostic significance of these variables. The final set of predictors derived from this process was used to construct the CS model. The nomogram was then established to visually demonstrate the relative contributions of the selected prognostic factors and to generate individualized, time-dependent survival estimates. Based on the risk scores calculated from the nomogram, patients were further stratified into distinct risk groups to facilitate risk differentiation. Model performance was systematically evaluated in both training and validation cohorts. Discrimination was assessed using the time-dependent receiver operating characteristic (ROC) curve and the concordance index (C-index). Calibration was examined by comparing predicted survival probabilities with observed outcomes at different time intervals using calibration plots. Clinical utility was assessed with decision curve analysis (DCA) to quantify the net benefit of the nomogram. Internal validation was performed through bootstrap resampling to evaluate the robustness and generalizability of the model.

Statistical analyses were carried out using R software. A two-sided P value below 0.05 was regarded as statistically significant, and 95% confidence intervals (CIs) were applied for all estimates.

Results

Incidence trends

From 2000 to 2021, the age-adjusted incidence rate of the disease remained relatively stable, ranging from 0.0541 to 0.0536 per 100,000 persons (Table 1 and Figure 1). The lowest rate was observed in 2018 (0.0399; 95% CI: 0.0282–0.0550), whereas the highest was recorded in 2014 (0.0741; 95% CI: 0.0570–0.0947, Figure 1). Overall, no consistent upward or downward trend was observed across the study period. Joinpoint regression analysis indicated an annual percent change (APC) of −0.16%, which was not statistically significant, suggesting that the incidence remained largely unchanged over the past two decades.

Figure 1 Age-adjusted incidence trends of primary breast diffuse large B-cell lymphoma in the United States from 2000 to 2021. The incidence remained relatively stable throughout the study period, with an estimated APC of –0.16% (not statistically significant). APC, annual percent change.

Patient demographics and tumor characteristics

The study cohort included 792 patients, with 554 in the training set and 238 in the validation set. Most patients were aged 60–79 years (46.2%) and White (77.3%). Early-stage disease (I/II) accounted for 62.1% of cases. Regarding treatment, 23.5% underwent surgery, 39.4% received radiotherapy, and 77.1% received chemotherapy. The majority of patients were married (50.4%) and lived in counties with median household incomes under $90,000 (59.0%). Distributions of demographic and clinical characteristics were similar between the training and validation cohorts (Table 2).

CS trends

The 10-year CS probabilities for the cohort are presented in Figure 2. At the time of diagnosis [0], the 1-year CS was 87%, gradually decreasing to 49% at 10 years. Notably, patients who had already survived 1 year demonstrated improved subsequent survival, with a 10-year CS of 57%, compared to the 49% from diagnosis. This trend persisted across subsequent landmark time points: for instance, the 10-year CS for patients who had survived 3 years was 65%, and for those who survived 5 years it reached 73%. These results indicated that the probability of surviving additional years increases for patients who have already survived the initial period after diagnosis, reflecting the dynamic improvement in prognosis over time.

Figure 2 Ten-year CS probabilities for patients with primary breast diffuse large B-cell lymphoma. CS estimates increased with longer survival after diagnosis, indicating improved long-term prognosis for patients who had already survived the initial years. CS, conditional survival.

CS nomogram construction and evaluation

In the training cohort, BSR initially identified age, disease stage, surgery, chemotherapy, and household income as the most informative prognostic variables (Figure 3). Stepwise backward Cox regression was then applied to simplify the model while maintaining strong predictive performance, ultimately retaining age, stage, and chemotherapy as the final predictors. Multivariable Cox analysis further confirmed the independent prognostic significance of these variables (Table 3). These factors were subsequently incorporated into the CS-nomogram (Figure 4). Using the nomogram, a risk score was calculated for each patient, and an optimal cutoff value of 72 was applied to stratify patients into high- and low-risk groups. This stratification effectively distinguished prognostic subgroups in both the training and validation cohorts, demonstrating the nomogram’s ability to dynamically discriminate patients with differing survival outcomes (Figure 5).

Figure 3 BSR analysis for variable selection in primary breast diffuse large B-cell lymphoma. BSR identified age, disease stage, surgery, chemotherapy, and household income as the most informative prognostic variables. BSR, best subset regression.

Figure 4 CS nomogram for patients with primary breast diffuse large B-cell lymphoma. CS, conditional survival; CT, chemotherapy; OS, overall survival.

Figure 5 Risk stratification of patients with primary breast diffuse large B-cell lymphoma based on the CS-nomogram. Using the optimal cutoff value of 72, patients were classified into high- and low-risk groups. Kaplan-Meier curves demonstrate clear separation of survival outcomes in both the training (A) and validation (B) cohorts. CS, conditional survival.

The performance of the CS nomogram was comprehensively evaluated in both the training and validation cohorts. Calibration plots demonstrated close agreement between the predicted survival probabilities and the observed outcomes across multiple time points, indicating high accuracy of the model (Figure 6A,6B). Discrimination of the nomogram was evaluated using the C-index and time-dependent ROC analysis, which demonstrated favorable predictive performance (Figure 6C,6D). The model showed good discriminative ability, with a C-index of 0.760 in the training cohort and 0.727 in the validation cohort, indicating its capability to distinguish patients with different survival outcomes. In the training cohort, the AUCs for 3-, 5-, and 10-year survival were 0.80, 0.79, and 0.83, respectively; in the validation cohort, the corresponding AUCs were 0.71, 0.77, and 0.80, reflecting robust predictive performance across multiple time points. Additionally, the favorable net benefit demonstrated by the DCA, together with the robust C-index and AUC values, further underscores the nomogram’s strong predictive and clinical applicability in PB-DLBCL. These findings indicate that the model provides meaningful clinical benefit across a broad range of threshold probabilities, supporting its potential utility in guiding individualized clinical decision-making (Figure 7A,7B).

Figure 6 Validation of the CS-nomogram in patients with primary breast diffuse large B-cell lymphoma. Calibration curves at 3-, 5-, and 10-year intervals in the training (A) and validation (B) cohorts, showing close agreement between predicted and observed survival probabilities. Time-dependent receiver operating characteristic curves for the training (C) and validation (D) cohorts, confirming the discriminative ability of the nomogram across different time points. CS, conditional survival.

Figure 7 DCA of the CS-nomogram for primary breast diffuses large B-cell lymphoma. DCA in the training (A) and validation (B) cohorts, demonstrating that the nomogram provides a greater net clinical benefit across a wide range of threshold probabilities. CS, conditional survival; DCA, decision curve analysis.

Discussion

Our study provided a comprehensive analysis of PB-DLBCL using data from the SEER database spanning from 2000 to 2021. We observed a stable age-adjusted incidence rate over the past two decades, with no significant upward or downward trend. CS analysis revealed that patients who survived the initial period post-diagnosis exhibited improved long-term survival probabilities. Furthermore, age, disease stage, and chemotherapy were identified as the optimal combination of prognostic factors through BSR followed by stepwise backward Cox regression analysis. These variables were subsequently incorporated into a CS-nomogram, which demonstrated robust performance in both training and validation cohorts.

The stable incidence rate of PB-DLBCL observed in our study aligns with previous reports indicating that primary breast lymphoma constitutes a small fraction of both breast malignancies and non-Hodgkin lymphomas. Despite its rarity, the consistent incidence underscored the importance of continued surveillance and research into this unique lymphoma subtype. Clinically, the low incidence may contribute to limited awareness and potential delays in diagnosis, emphasizing the need for heightened vigilance among clinicians when evaluating breast masses, particularly in older female patients.

Our CS analysis demonstrated that patients who survived the initial period post-diagnosis had progressively improved long-term survival probabilities. This finding was consistent with existing literature on CS in lymphoma patients, which suggested that survival probabilities increased over time for those who remain event-free. The findings enable more precise prognostic counseling and support tailored follow-up and treatment strategies for patients who have passed key post-diagnosis milestones.

Through BSR followed by stepwise backward Cox analysis, we identified age, disease stage, and chemotherapy as the optimal combination of prognostic factors for PB-DLBCL. Age at diagnosis emerged as a significant predictor of survival, consistent with previous reports in both primary breast and other extranodal or nodal sites of DLBCL (13). Disease stage at diagnosis, as classified by the Ann Arbor system, was confirmed as a significant prognostic factor in our cohort. Patients presenting with advanced-stage disease (III/IV) exhibited poorer survival compared with those with early-stage (I/II) disease, reflecting the greater tumor burden and potential for systemic dissemination. This observation aligned with prior studies in both primary breast and other extranodal DLBCL, emphasizing that stage remains a critical determinant of prognosis despite advances in therapy (3,13). The optimal therapeutic strategies for PB-DLBCL remain incompletely defined due to the rarity of this disease (14,15). In primary breast cancer, the development of modern chemotherapeutic strategies has firmly established chemotherapy as a cornerstone of treatment (16). Similarly, in our PB-DLBCL cohort, chemotherapy emerged as the most influential treatment-related factor, likely reflecting the widespread incorporation of rituximab as a standard component of chemoimmunotherapy (14,17,18). Previous studies had also shown that rituximab can effectively overcome PRDM1-associated resistance to chemotherapy (19,20), enhancing treatment efficacy. However, other investigations had reported that chemotherapy did not significantly improve OS or progression-free survival (PFS) in some PB-DLBCL cohorts (21-23), highlighting the heterogeneity of treatment response and the need for individualized therapy. Additionally, in the multivariable regression analysis, we observed a potential prognostic benefit of radiotherapy; however, the effect was not sufficiently strong to be retained in the final model. The role of RT in PB-DLBCL remains uncertain and warrants further study. Although some reports suggested a survival benefit (13,17,24,25), Chen et al. observed no significant improvement in OS (P=0.110), which may reflect unfavorable baseline characteristics among patients receiving RT in their cohort (26). Additionally, RT raises concerns about late toxicities, given that the treatment field often includes critical organs such as the heart and lungs (26). Another noteworthy finding was that surgery did not confer a survival benefit in our cohort. Consistent with current consensus, mastectomy appears to offer no clear advantage for patients with primary breast lymphoma (2,14,27), reinforcing the preference for systemic therapy over aggressive surgical intervention. Ultimately, integrating these key variables into our CS-nomogram offers a practical tool for individualized risk assessment and dynamic outcome prediction.

In the broader context of lymphoma progression, rare but clinically significant complications such as leptomeningeal disease (LMD) have gained increasing attention as OS improves in many cancers, including hematologic malignancies. Although more commonly reported in breast cancer, lung cancer, melanoma, and primary central nervous system (CNS) tumors, LMD shares similar mechanisms of CNS infiltration across cancer types, typically involving leptomeningeal invasion and cerebrospinal fluid dissemination (28). Diagnosis relies on cerebrospinal fluid cytology complemented by imaging modalities such as MRI or PET-CT. Despite its rarity, the growing recognition of LMD highlights the need for continued CNS-focused surveillance in aggressive lymphomas (28). While PB-DLBCL seldom presents with CNS involvement, insights from LMD research emphasize the importance of considering CNS risk when evaluating long-term disease progression in high-risk lymphoma subtypes.

The CS-nomogram developed in our study provides a user-friendly tool for clinicians to estimate individualized survival probabilities at multiple time points post-diagnosis. Its performance was rigorously evaluated using calibration plots, time-dependent ROC curves, and DCA. Calibration plots confirmed the agreement between predicted and observed outcomes, while ROC curves demonstrated good discriminative ability. Importantly, DCA highlighted the clinical utility of the model by quantifying the net benefit across a range of threshold probabilities, showing that using the nomogram to guide clinical decisions could improve patient outcomes compared with default strategies. These results are consistent with previous DLBCL nomogram studies (29-32), but our model offers the novel advantage of CS prediction, emphasizing its potential to further personalize follow-up and treatment strategies for patients at different post-diagnosis milestones.

There are several limitations in this study. The retrospective nature of the SEER database may introduce selection biases, and reliance on coding practices could affect the accuracy of treatment-related variables, including chemotherapy and radiotherapy details. Additionally, the absence of molecular and genetic data limits the integration of these factors into our prognostic model. Future studies incorporating genomic profiling, as well as more detailed prospective treatment data, are warranted to validate and refine our findings.

Conclusions

This study has provided valuable insights into the epidemiology and prognosis of PB-DLBCL. The development of a CS-nomogram offers a practical tool for clinicians to make informed decisions regarding patient management. Despite its limitations, our research contributes to the growing body of knowledge on this rare lymphoma subtype and underscores the importance of personalized approaches in oncology.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1976/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1976/prf

Funding: This work was supported by the Wenzhou Science and Technology Bureau (No. Y2023910), the Shangda Translational Medicine Fund under Grant (No. SDTMF2023KP02), and the China Foundation for Youth Entrepreneurship and Employment (No. CFYEE).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1976/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Aviv A, Tadmor T, Polliack A. Primary diffuse large B-cell lymphoma of the breast: looking at pathogenesis, clinical issues and therapeutic options. Ann Oncol 2013;24:2236-44. [Crossref] [PubMed]
2. Ryan G, Martinelli G, Kuper-Hommel M, et al. Primary diffuse large B-cell lymphoma of the breast: prognostic factors and outcomes of a study by the International Extranodal Lymphoma Study Group. Ann Oncol 2008;19:233-41. [Crossref] [PubMed]
3. Sun Y, Joks M, Xu LM, et al. Diffuse large B-cell lymphoma of the breast: prognostic factors and treatment outcomes. Onco Targets Ther 2016;9:2069-80. [Crossref] [PubMed]
4. Zuo H, Xie X, Sun X, et al. Prognostic impact of concordant and discordant bone marrow involvement on diffuse large B-cell lymphoma. Transl Cancer Res 2024;13:5339-46. [Crossref] [PubMed]
5. Zhao J, Gao J, Liu W, et al. Dynamic conditional survival nomogram for primary hepatocellular carcinoma: a population-based analysis. Discov Oncol 2025;16:854. [Crossref] [PubMed]
6. Ren X, Cong F, Chao G, et al. Individualized estimation of conditional survival for patients with spinal chordoma. Transl Cancer Res 2025;14:1710-24. [Crossref] [PubMed]
7. Zhang Y, Zhang M, Yu G, et al. Development and Validation of a Novel Conditional Survival Nomogram for Predicting Real-Time Prognosis in Patients With Breast Cancer Brain Metastasis. Clin Breast Cancer 2025;25:141-148.e1. [Crossref] [PubMed]
8. Meng X, Chang X, Qin P, et al. Risk-dependent conditional survival analysis and annual hazard rate of inflammatory breast cancer. Eur J Surg Oncol 2023;49:106957. [Crossref] [PubMed]
9. Zabor EC, Gonen M, Chapman PB, et al. Dynamic prognostication using conditional survival estimates. Cancer 2013;119:3589-92. [Crossref] [PubMed]
10. Kwong YD, Mehta KM, Miaskowski C, et al. Using best subset regression to identify clinical characteristics and biomarkers associated with sepsis-associated acute kidney injury. Am J Physiol Renal Physiol 2020;319:F979-87. [Crossref] [PubMed]
11. Zhang Z. Variable selection with stepwise and best subset approaches. Ann Transl Med 2016;4:136. [Crossref] [PubMed]
12. Matsumoto YK, Himoto Y, Nishio M, et al. Nodal infiltration in endometrial cancer: a prediction model using best subset regression. Eur Radiol 2024;34:3375-84. [Crossref] [PubMed]
13. Liu PP, Wang KF, Jin JT, et al. Role of radiation therapy in primary breast diffuse large B-cell lymphoma in the Rituximab era: a SEER database analysis. Cancer Med 2018;7:1845-51. [Crossref] [PubMed]
14. Franco Pérez F, Lavernia J, Aguiar-Bujanda D, et al. Primary Breast Lymphoma: Analysis of 55 Cases of the Spanish Lymphoma Oncology Group. Clin Lymphoma Myeloma Leuk 2017;17:186-91. [Crossref] [PubMed]
15. Yhim HY, Yoon DH, Kim SJ, et al. First-Line Treatment for Primary Breast Diffuse Large B-Cell Lymphoma Using Immunochemotherapy and Central Nervous System Prophylaxis: A Multicenter Phase 2 Trial. Cancers (Basel) 2020;12:2192. [Crossref] [PubMed]
16. Shah N, Mohammad AS, Saralkar P, et al. Investigational chemotherapy and novel pharmacokinetic mechanisms for the treatment of breast cancer brain metastases. Pharmacol Res 2018;132:47-68. [Crossref] [PubMed]
17. Hu S, Song Y, Sun X, et al. Primary breast diffuse large B-cell lymphoma in the rituximab era: Therapeutic strategies and patterns of failure. Cancer Sci 2018;109:3943-52. [Crossref] [PubMed]
18. Hariharan N, Kabadi A, Don M, et al. Composite Angioimmunoblastic T-Cell Lymphoma and Diffuse Large B-Cell Lymphoma Presenting with Distributive Shock. Hematol Rep 2023;15:627-33. [Crossref] [PubMed]
19. Liu YY, Leboeuf C, Shi JY, et al. Rituximab plus CHOP (R-CHOP) overcomes PRDM1-associated resistance to chemotherapy in patients with diffuse large B-cell lymphoma. Blood 2007;110:339-44. [Crossref] [PubMed]
20. Franco F, González-Rincón J, Lavernia J, et al. Mutational profile of primary breast diffuse large B-cell lymphoma. Oncotarget 2017;8:102888-97. [Crossref] [PubMed]
21. Zhang N, Cao C, Zhu Y, et al. Primary breast diffuse large B-cell lymphoma in the era of rituximab. Onco Targets Ther 2016;9:6093-7. [Crossref] [PubMed]
22. Aviles A, Neri N, Nambo MJ. The role of genotype in 104 cases of diffuse large B-cell lymphoma primary of breast. Am J Clin Oncol 2012;35:126-9. [Crossref] [PubMed]
23. Hosein PJ, Maragulia JC, Salzberg MP, et al. A multicentre study of primary breast diffuse large B-cell lymphoma in the rituximab era. Br J Haematol 2014;165:358-63. [Crossref] [PubMed]
24. Haque W, Dabaja B, Tann A, et al. Changes in treatment patterns and impact of radiotherapy for early stage diffuse large B cell lymphoma after Rituximab: A population-based analysis. Radiother Oncol 2016;120:150-5. [Crossref] [PubMed]
25. Weng H, Shrestha PR, Hong H, et al. Primary breast diffuse large B-cell lymphoma in the rituximab era: A retrospective study of the Chinese Southwest Oncology Group. Cancer Med 2023;12:21188-98. [Crossref] [PubMed]
26. Chen GL, Li D, Cao S, et al. Clinical Characteristics and Prognostic Factors in Primary Breast Diffuse Large B-Cell Lymphoma. Mediterr J Hematol Infect Dis 2022;14:e2022066. [Crossref] [PubMed]
27. Jennings WC, Baker RS, Murray SS, et al. Primary breast lymphoma: the role of mastectomy and the importance of lymph node status. Ann Surg 2007;245:784-9. [Crossref] [PubMed]
28. Nguyen A, Nguyen A, Dada OT, et al. Leptomeningeal Metastasis: A Review of the Pathophysiology, Diagnostic Methodology, and Therapeutic Landscape. Curr Oncol 2023;30:5906-31. [Crossref] [PubMed]
29. Li Z, Shi J, Wu X, et al. Machine learning-based histopathological features of histological slides and clinical characteristics as a novel prognostic indicator in diffuse large B-cell lymphoma. Pathol Res Pract 2025;272:156071. [Crossref] [PubMed]
30. Liu Y, Han H, Wei H, et al. Predictive Modelling of Overall Survival in Adult Patients with Primary Diffuse Large B-cell Lymphoma of the Breast Using the Surveillance, Epidemiology, and End Results (SEER) Database. Recent Pat Anticancer Drug Discov 2024;19:373-82. [Crossref] [PubMed]
31. Lang M, Feng Y, Meng X, et al. Improved method to stratify lymphoma patients with risk of secondary central nervous system involvement: A multicenter retrospective analysis. Hematol Oncol 2023;41:239-47. [Crossref] [PubMed]
32. Peng X, Yu S, Kou Y, et al. Prediction nomogram based on 18F-FDG PET/CT and clinical parameters for patients with diffuse large B-cell lymphoma. Ann Hematol 2023;102:3115-24. [Crossref] [PubMed]