Establishment and evaluation of a nomogram for predicting the survival outcomes of patients with diffuse large B-cell lymphoma based on International Prognostic Index scores and clinical indicators

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma (NHL) and accounts for 30–40% of all new diagnoses of NHL (1). Patients with DLBCL differ in terms of their morphologic, immunophenotypic, clinical, and genetic features, and this heterogeneity underlies their different responses to chemotherapy regimens (2). Previously, the standard first-line treatment was cyclophosphamide, adriamycin, vincristine, and prednisone (CHOP), but more recent studies have shown that the addition of rituximab to CHOP (R-CHOP) decreases the risk of early and late recurrence (3,4). Subsequent clinical study reported that R-CHOP cures >50% of DLBCL patients, and this regimen is now accepted as a standard first-line treatment for DLBCL (5). However, some patients cannot receive rituximab treatment for various reasons, such as economic limitations and/or disease status, which may affect the effectiveness of treatments and patient prognosis. Thus, the use of rituximab should be considered when assessing the prognosis of patients with DLBCL.

The precise prognostic evaluation of patients with DLBCL is important for developing intervention and management strategies. Previous studies have used the International Prognostic Index (IPI) for the prognostic stratification of DLBCL patients. The IPI considers age, Ann Arbor stage, Eastern Cooperative Oncology Group Performance Status (ECOG PS), extra-nodal disease sites, and the serum level of lactate dehydrogenase (LDH). Subsequent studies have attempted to improve predictions of survival by revising this scale. For example, in 2007 Sehn et al. (6) established a revised-IPI (R-IPI) score system for patients who received the R-CHOP regimen. This new system categorized patients into 3 groups, was easier to use, and performed slightly better than the original IPI system. In 2014, Zhou et al. (7) established the National Comprehensive Cancer Network IPI (NCCN-IPI) score system. This new system assigned different weights to variables of age and LDH and had high sensitivity in the identification of patients with a 5-year survival probability rate <50%. However, these IPI-based scoring systems remain controversial, as some studies have found their predictive ability is not optimal for DLBCL patients who were identified as having a poor risk of survival (8,9).

With the continued development of new therapeutic methods and testing techniques, the choice of treatment regimens and potential clinical indexes should be reconsidered, as these differences may be responsible for the heterogeneity in the prognoses of patients with DLBCL. In an effort to improve the predictive performance of the IPI scoring system, we established and validated prognostic models that integrate specific clinical variables with the IPI to provide individualized survival risk assessments for patients with DLBCL. We present the following article in accordance with the TRIPOD reporting checklist (available at https://atm.amegroups.com/article/view/10.21037/atm-22-6023/rc).

Methods

Study population and data source

Patients diagnosed with DLBCL between January 2008 and December 2017 at the Harbin Medical University (HMU) Cancer Hospital (Heilongjiang, China) were enrolled in this study, the follow-up cut-off date was 31 August 2020. All the patients were classified using the 2008 World Health Organization’s Classification of Tumors of Hematopoietic and Lymphoid Tissue. Patients were excluded if they were aged <18 or >80 years old, did not have complete clinical information, had a primary central nervous system (CNS) lymphoma, had positive serological status for human immunodeficiency virus (HIV), or received first-line immunochemotherapy without a CHOP-like or R-CHOP-like regimen. Table 1 shows the rationales for these exclusion criteria. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This study was approved by the Ethics Committee of the HMU Cancer Hospital (approval No. GZR2014-069). The individuals in our study signed informed consent forms.

Definition of variables

The primary event for the overall survival (OS) analysis was death. The time of OS and other outcome measures were defined as the time from the day of the DLBCL diagnosis until the event of interest. Progression-free survival (PFS) was defined as the time from the beginning of treatment until disease progression, recurrence, or death. As of the last follow-up, there were 150 deaths and 270 survivors. The following baseline demographic and clinical variables were recorded: age (18–59 or ≥60 years), gender (male or female), B symptoms (absent or present), pathological subtype [germinal center B (GCB) or non-GCB], IPI risk stratification group (low, low-intermediate, high-intermediate, or high), first-line treatment regimen (CHOP-like or R-CHOP-like), lymphocyte-monocyte ratio (LMR; ≤3 or >3), neutrophil-lymphocyte ratio (NLR; ≤2.5 or >2.5), and Ki-67 index (≤70% or >70%). The IPI score system considers age, Ann Arbor stage, extra-nodal disease sites, and ECOG PS.

Statistical analysis

The optimal cut-off values for the LMR (3) and NLR (2.5) were obtained by analyzing the significance of the log-rank results using a minimal P value approach with Cutoff Finder (10). The Kaplan-Meier method was used to analyze OS, and survival differences among the groups were compared using the log-rank test. The multivariate survival analysis included all the variables that were significant in the univariate analysis, and Cox proportional hazards regression was used to calculate hazard ratios (HRs) and 95% confidence interval (CIs). The nomogram for outcomes at 3 years and 5 years were constructed using significant risk factors from the multivariate Cox regression model. The Harrell concordance index (C-index) was used to assess the accuracy of the nomogram from 1000 bootstrap re-samples. The c-index is considered acceptable between 0.7 and 0.8, is considered excellent 0.8 and 0.9, and more than 0.9 is considered outstanding. Calibration curves were applied for the model validation, and points closer to the diagonal line (i.e., predictions matching observations) indicated a more accurate model. A decision curve analysis (DCA) was used to estimate the clinical usefulness and net benefit of the nomograms. All the statistical tests were 2-sided, and a P value <0.05 was considered significant. The statistical analysis was performed using R software version 3.6.1 (R Foundation, Vienna, Austria). The R package survminer function ggsurvplot was used to draw the survival curves.

Results

Patient characteristics and survival

We retrospectively examined the records of 420 DLBCL patients for whom active follow-up data were available (Table 2). The patients had a median age of 55 years, and were approximately equal in terms of the number of males and females (209 males, 211 females). A total of 100 patients (23.8%) had B symptoms and 291 (69.3%) had the non-GCB pathological subtype. The IPI classification indicated that 234 (55.7%) had a low risk, 85 (20.2%) had a low-intermediate risk, 58 (13.8%) had a high-intermediate risk, and 43 (10.2%) had a high risk. The first-line treatment was CHOP-like in 234 patients (55.7%) and R-CHOP-like in 186 patients (44.3%). A total of 242 patients (57.6%) had a LMR >3, 270 (64.3%) had a Ki-67 value >70%, and 201 (47.9%) had a NLR >2.5. The median PFS time was 110 months, and the median OS time was not reached; the 3- and 5-year OS rates were 76.18% and 70.62%, respectively, and the 3- and 5-year PFS rates were 63% and 59.02%, respectively.

Univariate and multivariate analyses

We performed univariate and multivariate Cox proportional hazards regression analyses to identify the potential prognostic factors for OS (Figure 1 and Table 3) and PFS (Figure 2 and Table 4). The univariate Cox regression analysis indicated that age at the time of diagnosis (Figure 1A, P=0.008), B symptoms (Figure 1B, P<0.001), the LMR (Figure 1C, P<0.001), the NLR (Figure 1D, P=0.002), the IPI (Figure 1E, P<0.001), the use of a first-line regimen with rituximab (Figure 1F, P=0.04), and Ki-67 (Figure 1G, P=0.002) were significantly associated with OS. Similarly, pathological subtype (Figure 2A, P=0.006), B symptoms (Figure 2B, P<0.001), the LMR (Figure 2C, P<0.001), the NLR (Figure 2D, P=0.002), the IPI (Figure 2E, P<0.001), the use of a first-line regimen with rituximab (Figure 2F, P=0.04), and Ki-67 (Figure 2G, P=0.002) were associated with PFS. The multivariate Cox regression model, which included the aforementioned significant factors, showed that a low IPI score, the use of a first-line regimen with rituximab, a high LMR, and a Ki-67 <70% were significantly associated with improved OS and PFS (all P<0.01).

Figure 1 Kaplan-Meier curves of OS in patients with stratification by age (A), B symptoms (B), LMR (C), NLR (D), IPI (E), chemotherapy regimen (F), and Ki-67 (G). OS, overall survival; LMR, lymphocyte-to-monocyte ratio; NLR, neutrophil-to-lymphocyte ratio; IPI, International Prognostic Index; R-CHOP, rituximab plus cyclophosphamide, adriamycin, vincristine, and prednisone.

Figure 2 Kaplan-Meier curves of PFS in patients with stratification by pathological subtypes (A), B symptoms (B), LMR (C), NLR (D), IPI (E), chemotherapy regimen (F), and Ki-67 (G). PFS, progression-free survival; LMR, lymphocyte-to-monocyte ratio; NLR, neutrophil-to-lymphocyte ratio; IPI, International Prognostic Index; R-CHOP, rituximab plus cyclophosphamide, adriamycin, vincristine, and prednisone.

Survival prediction model

We used the results of the multivariate analyses to construct nomograms for OS (Figure 3A) and PFS (Figure 3B) based on the 4 factors that were significantly associated with the outcomes in the multivariate analysis. These nomograms had good accuracy for predicting OS (C-index =0.73) and PFS (C-index =0.68), and slightly outperformed the IPI (OS: C-index =0.67; PFS: C-index =0.63). The survival probability calibration curves for OS and PFS showed good agreement between the actual survival probability and nomogram-predicted survival probability (Figure 3C-3F). The DCA results showed that the net benefit of our model was greater than the screening using the original IPI system across a wide range of thresholds (0.01–0.86; Figure 3G,3H).

Figure 3 Nomograms for predicting 3-year and 5-years OS (A) and PFS (B). Calibration curves for 3-year and 5-year OS (C,E) and PFS (D,F). DCA for 5-year OS (G) and PFS (H). OS, overall survival; PFS, progression-free survival; DCA, decision curve analysis.

The Kaplan-Meier survival curves based on risk stratification using the modified IPI (Figure 4A,4B) were better able to discriminate among patients with different prognoses than survival curves using the IPI (Figure 1E,2E).

Figure 4 Kaplan-Meier curves of OS and PFS with risk stratification based on the modified IPI. See Figures 1E,2E for comparison. OS, overall survival; PFS, progression-free survival; IPI, International Prognostic Index.

Discussion

We developed and validated nomograms that combined clinical prognostic factors with the IPI to predict the 3-year and 5-year OS and PFS of DLBCL patients who received standard first-line treatment regimens. The study has reported that the IPI showed consistently good performance in risk stratification and prognostic prediction, and that the R-IPI and age-adjusted (aa)-IPI provided no substantial improvements (11). Following the clinical introduction of rituximab, the original intent of the NCCN-IPI was to identify the subgroup of high-risk patients with poor survival by stratification using risk factors based on the IPI, and to re-assign scores based on different prognostic risk factors. The results showed that the NCCN-IPI discriminated among patients at poor risk using PFS instead of OS as the end point, but failed to identify the prognosis of patients with poor outcomes (12). Because the results of the IPI and NCCN-IPI did not differ significantly in terms of treatment decisions, and the IPI was simpler and easier to use, we based our prediction model on the IPI score system.

The previous IPI and NCCN-IPI risk scoring systems did not consider the effects of rituximab treatment. In our study, we enrolled patients receiving standard first-line treatment either with or without rituximab. Our results showed that the use of rituximab provided a significant survival advantage. The initial combination of rituximab with chemotherapy has been shown to reduce the risk of relapse and improve PFS and long-term OS, rituximab has also been shown to be more effective in younger patients with a good prognosis than elderly patients (13). The selection of treatment regimen relies on a physician’s discretion considering the specific status of each patient. The study reported that standard-dose rituximab (375 mg/m²) significantly increased the risk of HBV reactivation compared with reduced-dose rituximab (200 mg/body) and no rituximab (14). The research also has found that rituximab is implicated in precipitating tumor lysis syndrome (TLS) in patients with high tumor burden (15). Clinicians recommend rituximab to patients who are in remission for comorbid conditions and who have clinical indication for rituximab therapy. Thus, some patients will not receive a regimen with rituximab for a variety of reasons as the first-line therapy. The nomograms presented in this study provide a visualization of the survival benefit of using rituximab and could be used to guide patient stratification and assist in the making of reasonable therapeutic decisions. In the absence of rituximab therapy, the follow-up strategies and management must be reconsidered; for example, adjusting the dosage and/or intensity of the chemotherapy regimen may improve efficacy (16).

Numerous factors affect oncogenesis, tumor progression, response to therapy, and the outcome of patients with lymphoma, including the tumor microenvironment, host immunity, and inflammatory responses (17). The LMR prior to the initiation of the first-line treatment is an important indicator of DLBCL prognosis, and a lower LMR is associated with a poor prognosis because it indicates poor host immunity, a factor that contributes to lymphoma progression (18). This is consistent with our results, which indicated that an LMR <3 is significantly associated with a reduced OS and PFS. Lymphocytes are important mediators of antibody-dependent cell-mediated cytotoxicity, and lymphopenia is an indicator of poor prognosis in DLBCL due to the reduced tumor clearance and surveillance ability of the host (19). An increased level of monocytes in the peripheral blood can promote tumor progression by increasing angiogenesis, inhibit anti-tumor immunity, and increase the proliferation of malignant cells (20). A low LMR favors tumorigenesis and progression by increasing the levels of cytokines released by tumor cells, such as granulocyte colony-stimulating factor. These cytokines promote the proliferation and differentiation of neutrophils, which function in the microenvironment required for tumor development, induce tumor angiogenesis, and thus promote the proliferation, invasion, and metastasis of tumor cells (21).

Previous research has reported that a Ki-67 cut-off value of 70% can be used to reliably discriminate between DLBCL patients with good and poor prognoses (22). Similarly, our results showed that a Ki-67 index >70% indicated worse OS and PFS. Other researches have also shown that a high Ki-67 index is associated with survival in patients with NHL, such as DLBCL, mantle cell lymphoma (23), and natural killer/T cell lymphoma (24). However, since the introduction of rituximab, a high Ki-67 index in DLBCL patients has been considered an indicator of a favorable prognosis (25). High Ki-67 expression might be indicative of a tumor with an elevated risk for regeneration and mutations, leading to treatment failure, or active proliferation, meaning that the tumor is more susceptible to treatment (26). In our view, it is important to consider differences in how Ki-67 is measured at different medical institutions (e.g., the staining method used to determine the proportion of positive Ki-67 cells). In this study, all the pathology reports were re-evaluated by a pathologist at our institution, and Ki-67 was included as a factor in our nomograms.

The present study differs from previous studies because we established and validated nomograms for use in a population of DLBCL patients who received standard first-line treatment regimens either with or without rituximab. We also considered the IPI, along with features of the tumor micro-environment and patient pathological characteristics, in developing our nomograms. Our approach provided improved predictions of patient survival and identified high-risk patients who may need novel or aggressive therapies or who may be suitable for enrolment in clinical trials that aim to prolong survival time.

Our study also had some limitations. First, it was conducted at a single-center with a relatively small number of patients, so validation is needed by a large multi-center study. Second, due to the retrospective nature of this study, we will continue to collect clinically relevant data for analysis at a later time. Third, there may be additional opportunities to improve the nomogram so that it better satisfies clinical requirements.

Conclusions

In conclusion, our study identified prognostic factors in patients who were newly diagnosed with DLBCL and constructed an individualized risk prediction model. We found that LMR, Ki-67, the use of rituximab in the first-line treatment, and IPI score were significantly associated with OS and PFS. Our modified IPI scoring system, might be a valuable tool that can be used to reliably predict. It enables clinicians to identify patients with possible poor prognosis and choose more active treatment for patients, so as to prolong the PFS and OS of patients.

Acknowledgments

Funding: This study was funded by the National Natural Science Foundation of China (No. 8187010189).

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://atm.amegroups.com/article/view/10.21037/atm-22-6023/rc

Data Sharing Statement: Available at https://atm.amegroups.com/article/view/10.21037/atm-22-6023/dss

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://atm.amegroups.com/article/view/10.21037/atm-22-6023/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 individuals in our study signed informed consent forms. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This study was approved by the Ethics Committee of the HMU Cancer Hospital (approval No. GZR2014-069).

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. Chaganti S, Illidge T, Barrington S, et al. Guidelines for the management of diffuse large B-cell lymphoma. Br J Haematol 2016;174:43-56. [Crossref] [PubMed]
2. Ennishi D, Hsi ED, Steidl C, et al. Toward a New Molecular Taxonomy of Diffuse Large B-cell Lymphoma. Cancer Discov 2020;10:1267-81. [Crossref] [PubMed]
3. Modvig L, Vase M, d'Amore F. Clinical and treatment-related features determining the risk of late relapse in patients with diffuse large B-cell lymphoma. Br J Haematol 2017;179:75-82. [Crossref] [PubMed]
4. Deng HG, Zhou YL, Wen J, et al. Development and validation of nomograms by radiomic features on ultrasound imaging for predicting overall survival in patients with primary nodal diffuse large B-cell lymphoma. Front Oncol 2022;12:991948. [Crossref] [PubMed]
5. Pfreundschuh M, Trümper L, Osterborg A, et al. CHOP-like chemotherapy plus rituximab versus CHOP-like chemotherapy alone in young patients with good-prognosis diffuse large-B-cell lymphoma: a randomised controlled trial by the MabThera International Trial (MInT) Group. Lancet Oncol 2006;7:379-91. [Crossref] [PubMed]
6. Sehn LH, Berry B, Chhanabhai M, et al. The revised International Prognostic Index (R-IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B-cell lymphoma treated with R-CHOP. Blood 2007;109:1857-61. [Crossref] [PubMed]
7. Zhou Z, Sehn LH, Rademaker AW, et al. An enhanced International Prognostic Index (NCCN-IPI) for patients with diffuse large B-cell lymphoma treated in the rituximab era. Blood 2014;123:837-42. [Crossref] [PubMed]
8. Ruppert AS, Dixon JG, Salles G, et al. International prognostic indices in diffuse large B-cell lymphoma: a comparison of IPI, R-IPI, and NCCN-IPI. Blood 2020;135:2041-8. [Crossref] [PubMed]
9. Song JL, Wei XL, Zhang YK, et al. The prognostic value of the international prognostic index, the national comprehensive cancer network IPI and the age-adjusted IPI in diffuse large B cell lymphoma. Zhonghua Xue Ye Xue Za Zhi 2018;39:739-44. [PubMed]
10. Budczies J, Klauschen F, Sinn BV, et al. Cutoff Finder: a comprehensive and straightforward Web application enabling rapid biomarker cutoff optimization. PLoS One 2012;7:e51862. [Crossref] [PubMed]
11. Ziepert M, Hasenclever D, Kuhnt E, et al. Standard International prognostic index remains a valid predictor of outcome for patients with aggressive CD20+ B-cell lymphoma in the rituximab era. J Clin Oncol 2010;28:2373-80. [Crossref] [PubMed]
12. Gleeson M, Counsell N, Cunningham D, et al. Prognostic indices in diffuse large B-cell lymphoma in the rituximab era: an analysis of the UK National Cancer Research Institute R-CHOP 14 versus 21 phase 3 trial. Br J Haematol 2021;192:1015-9. [Crossref] [PubMed]
13. Shi Q, Schmitz N, Ou FS, et al. Progression-Free Survival as a Surrogate End Point for Overall Survival in First-Line Diffuse Large B-Cell Lymphoma: An Individual Patient-Level Analysis of Multiple Randomized Trials (SEAL). J Clin Oncol 2018;36:2593-602. [Crossref] [PubMed]
14. Lee JH, Park JY, Kim DG, et al. Effects of rituximab dose on hepatitis B reactivation in patients with resolved infection undergoing immunologic incompatible kidney transplantation. Sci Rep 2018;8:15629. [Crossref] [PubMed]
15. Yang B, Lu XC, Yu RL, et al. Diagnosis and Treatment of Rituximab-Induced Acute Tumor Lysis Syndrome in Patients With Diffuse Large B-Cell Lymphoma. AmJ Med Sci 2012;343:337-41. [Crossref] [PubMed]
16. Epperla N, Vaughn JL, Othus M, et al. Recent survival trends in diffuse large B-cell lymphoma--Have we made any progress beyond rituximab? Cancer Med 2020;9:5519-25. [Crossref] [PubMed]
17. Autio M, Leivonen SK, Brück O, et al. Immune cell constitution in the tumor microenvironment predicts the outcome in diffuse large B-cell lymphoma. Haematologica 2021;106:718-729. [Crossref] [PubMed]
18. Bento L, Díaz-López A, Barranco G, et al. New prognosis score including absolute lymphocyte/monocyte ratio, red blood cell distribution width and beta-2 microglobulin in patients with diffuse large B-cell lymphoma treated with R-CHOP: Spanish Lymphoma Group Experience (GELTAMO). Br J Haematol 2020;188:888-897. [Crossref] [PubMed]
19. Mato A, Feldman T, Zielonka T, et al. Rituximab, cyclophosphamide-fractionated, vincristine, doxorubicin and dexamethasone alternating with rituximab, methotrexate and cytarabine overcomes risk features associated with inferior outcomes in treatment of newly diagnosed, high-risk diffuse large B-cell lymphoma. Leuk Lymphoma 2013;54:2606-12. [Crossref] [PubMed]
20. Ma JJ, Pang XL, Li JN, et al. The immune checkpoint expression in the tumor immune microenvironment of DLBCL: Clinicopathologic features and prognosi. Front Oncol 2022;12:1069378. [Crossref] [PubMed]
21. Baecklund E, Smedby KE, Sutton LA, et al. Lymphoma development in patients with autoimmune and inflammatory disorders--what are the driving forces? Semin Cancer Biol 2014;24:61-70. [Crossref] [PubMed]
22. Broyde A, Boycov O, Strenov Y, et al. Role and prognostic significance of the Ki-67 index in non-Hodgkin's lymphoma. Am J Hematol 2009;84:338-43. [Crossref] [PubMed]
23. Sohani AR, Maurer MJ, Giri S, et al. Biomarkers for Risk Stratification in Patients With Previously Untreated Follicular Lymphoma Receiving Anti-CD20-based Biological Therapy. Am J Surg Pathol 2021;45:384-93. [Crossref] [PubMed]
24. Zeng LS, Huang WT, Cao Z, et al. The correlation of clinicopathological features and prognosis in extranodal natural killer/T cell lymphoma: a report of 42 cases in the early stage. Ann Hematol 2019;98:1467-76. [Crossref] [PubMed]
25. Salles G, de Jong D, Xie W, et al. Prognostic significance of immunohistochemical biomarkers in diffuse large B-cell lymphoma: a study from the Lunenburg Lymphoma Biomarker Consortium. Blood 2011;117:7070-8. [Crossref] [PubMed]
26. Andrés-Sánchez N, Fisher D, Krasinska L. Physiological functions and roles in cancer of the proliferation marker Ki-67. J Cell Sci 2022;135:jcs258932. [Crossref] [PubMed]

(English Language Editor: L. Huleatt)