Abstract
We recently defined event-free survival at 24 months (EFS24) as a clinically relevant outcome for patients with DLBCL. Patients who fail EFS24 have very poor overall survival, while those who achieve EFS24 have a subsequent overall survival equivalent to that of the age- and sex-matched general population. Here, we develop and validate a clinical risk calculator (IPI24) for EFS24. Model building was performed on a discovery dataset of 1,348 patients with DLBCL and treated with anthracycline-based immunochemotherapy. A multivariable model containing age, Ann Arbor stage, normalized serum LDH, ALC, ECOG performance status, bulky disease, and sex was identified. The model was then applied to an independent validation dataset of 1,177 DLBCL patients. The IPI24 score estimates the probability of failing to achieve the EFS24 endpoint for an individual patient. The IPI24 model showed superior discriminatory ability (c-statistic = 0.671) in the validation dataset compared to the IPI (c-statistic = 0.649) or the NCCN-IPI (c-statistic = 0.657). After recalibration of the model on the combined dataset, the median predicted probability of failing to achieve EFS24 was 36% (range, 12–88%), and the IPI24 showed an EFS24 gradient in all IPI groups. The IPI24 also identified a significant percentage of patients with high risk disease, with over 20% of patients having a 50% or higher risk of failing to achieve EFS24. The IPI24 provides an individual patient level probability of achieving the clinically relevant EFS24 endpoint. It can be used via electronic apps.
Introduction
Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of lymphoma in the western world. Standard initial therapy for DLBCL is rituximab plus anthracycline containing multiagent chemotherapy regimen, typically given as R-CHOP [1–3]. Nearly all patients respond to initial immunochemotherapy and R-CHOP, or comparable regimens are curative in the majority of patients. However, about one-third of patients will have refractory disease or relapse following R-CHOP therapy, and outcome is generally poor for these patients despite salvage therapies [4]. Therefore, risk stratification at diagnosis for patients with DLBCL is of great clinical interest.
We have recently shown that event-free survival status at 24 months (EFS24) is a robust endpoint for assessing disease-related outcome in patients with DLBCL treated in the rituximab era [5]. Patients who achieve EFS24 have a subsequent survival that is equivalent to the age- and sex-matched general population. In contrast, patients who have a relapse or retreatment event in the first 24 months from diagnosis have very poor, disease-related outcome with a median survival after event of 10 months. In addition to its clinical relevance and potential adoption as a clinical trial endpoint, EFS24 status is an appealing endpoint for a prognostic model in DLBCL. The IPI and its numerous updates have been built on time-to-event endpoints such overall or EFS. This results in the lumping of patients into a small number of risk groups whose outcome is given by a survival curve. While these survival models generally do a good job of stratifying outcome based on risk groups, the predicted outcomes at a specific timepoint, often given as survival at 5 years, can vary widely depending on age and clinical characteristics of the patients in the dataset. Thus, while a patient may be classified as “high risk” or “low risk,” it is difficult to council patients on their individual risks for disease specific outcomes. The dichotomous nature of the EFS24 endpoint allows an individual risk prediction, and the specificity of the EFS24 endpoint allows us to better model risk of disease related outcomes in a setting where the majority of patients are cured with front-line therapy.
Therefore, we performed the first study to assess the association of standard lymphoma clinical characteristics with the EFS24 endpoint. We then generated a personalized prognostic model (IPI24) for the EFS24 endpoint, validated its performance in an independent dataset, and demonstrate its superior performance to existing prognostic indices when applied to the EFS24 endpoint. We have also developed a nomogram and translated this to an easy-to-use smart phone application.
Methods
Patients for the study were prospectively enrolled in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER) [6–8], NCCTG N0489 clinical trial [9], ECOG E4494 clinical trial [2], a Lyon, France hospital registry, and the GELA LNH2003B clinical trial program [10–14]. Briefly, all patients had pathology confirmed, newly diagnosed DLBCL, and received initial therapy with immunochemotherapy or were enrolled in a randomized clinical trial where one of the arms contained immunochemotherapy. Patients with primary mediastinal B-cell lymphoma were included; patients with grey zone lymphoma, transformation of a previously diagnosed indolent lymphoma, primary central nervous system lymphoma, or post-transplant lymphoproliferative disorders were excluded. Patients enrolled on clinical trials were treated and followed per trial protocol. Patients on the MER were managed per treating physician and were contacted every 6 months during the first 3 years and annually thereafter to follow for disease progression/relapse, retreatment and death; all events were verified using medical records. Variables for analysis were abstracted from medical records or research material per clinical trial protocol or per standard MER abstraction procedures. This study was reviewed and approved by the human subjects institutional review board at the Mayo Clinic.
Discovery dataset
The discovery dataset consisted of patients included in a previously published study examining EFS24 in DLBCL [5], which included MER patients enrolled from 2002 to 2009, NCCTG N0489 patients, and GELA LNH2003B patients. 1,348 of the 1,587 patients (85%) in the previously published dataset were included. The 15% excluded were composed of patients from the Lyon cohort who were not enrolled in the GELA LNH2003B clinical trial program and had no data on continuous lactate dehydrogenase (LDH), absolute lymphocyte count (ALC), and bulky disease.
Replication dataset
An independent replication dataset was created using 270 patients enrolled in the MER through December 31, 2011 which were not included in the discovery dataset, 640 patients enrolled in the GELA 2003B program which were not included in the discovery dataset, and 267 patients treated on the R-CHOP arm of Eastern Cooperative Group (ECOG) E4494 [2].
Statistical methods
EFS was defined as time from diagnosis to relapse, retreatment after initial immunochemotherapy, or death due to any cause. Overall survival was defined as time from date of diagnosis to death due to any cause. EFS24 was defined as EFS status 24 months after date of diagnosis. Association of clinical characteristics with EFS24 was performed using logistic regression; odds ratios and c-statistics were used to summarize associations. Full details on variable assessment, model building, and model evaluation can be found in the Supporting Information material.
Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data and had final responsibility for the decision to submit for publication.
Results
Patient characteristics in the discovery dataset
The discovery dataset consisted of 1,348 patients with prognostic data available for modeling from our previously published study of several clinical trials and an observational cohort study [5]. The median age in the discovery dataset was 62 years (range 17–93), and 55% of patients were male (Supporting Information Table 4). Most patients (68%) had stage III–IV disease, and 58% had an elevated LDH. At a median follow-up of 59 months (range 1–135), 531 patients (39%) had an event, and 372 patients (28%) had died; 70% of patients achieved EFS24 (Kaplan-Meier estimate) with 30% failing to achieve EFS24.
Model building
Variables assessed for association with EFS24 were sex (male vs. female), age (≤60 vs. >60 years; continuous), stage (I, II, III, IV), extranodal sites (<2 vs. ≥2), NCCN-IPI-defined extranodal sites (i.e., disease involvement of central nervous system (CNS), gastrointestinal (GI) tract, lung, liver, bone marrow), serum LDH (<ULN vs. ≥upper limit of normal (ULN)), normalized serum LDH (LDH/ULN, continuous), Eastern Cooperative Group performance score (ECOG PS) (0, 1, 2, 3–4), bulky disease (<10 cm vs. ≥10 cm), bone marrow involvement (absent vs. present), B-symptoms (present vs. absent), presence of indolent lymphoma at time of initial diagnosis (i.e., composite histology), ALC (continuous value derived from blood counts), and cell-of-origin (COO) per Hans criteria (germinal center B-cell (GCB) vs. non-GCB) [15]. All variables showed an association with EFS24 in univariate analysis, with the exception of sex (male OR = 1.19, 95% CI: 0.95–1.50, P = 0.13), (Table I). The most discriminatory variables in univariate analysis were LDH (spline c-statistic = 0.660, P = 1.97 × 10−17), ECOG PS (multicategory c-statistic = 0.620, P = 4.94 × 10−16), ALC (spline c-statistic=0.608, P = 6.15 × 10−9), and stage (multicategory c-statistic=0.588, P = 1.38 × 10−12). Functional forms for continuous variables were examined, supporting a nonlinear effect for normalized LDH between the range of (0.5 and 5), a linear effect for age above 70 years, and a linear effect for ALC over the range of (0 and 2), (Supporting Information Fig. 1).
TABLE I.
Univariate Logistic Regression Models for EFS24 in the Discovery Dataset (N = 1,348)
Variable | % fail EFS24 | OR | Lower 95% CI | Upper 95% CI | Model p-value | C-Index |
---|---|---|---|---|---|---|
Sex | ||||||
Female | 31% | Ref | 1.34E – 01 | 0.516 | ||
Male | 35% | 1.19 | 0.95 | 1.50 | ||
Age | ||||||
≤60 | 29% | Ref | 4.91E – 04 | 0.531 | ||
>60 | 37% | 1.50 | 1.19 | 1.89 | ||
Continuous, spline | 6.67E – 05 | 0.562 | ||||
Stage (four levels) | ||||||
I | 12% | Ref | 1.38E – 12 | 0.588 | ||
II | 30% | 3.02 | 1.80 | 5.07 | ||
III | 39% | 4.53 | 2.69 | 7.62 | ||
IV | 39% | 4.60 | 2.89 | 7.33 | ||
Stage (two levels) | ||||||
I–II | 22% | Ref | 5.26E – 10 | 0.577 | ||
III–IV | 39% | 2.24 | 1.72 | 2.91 | ||
Number of extranodal sites | ||||||
0–1 | 30% | Ref | 4.39E – 04 | 0.540 | ||
≥2 | 41% | 1.56 | 1.22 | 2.00 | ||
Extranodal site (NCCN) | ||||||
No | 29% | Ref | 4.03E – 06 | 0.558 | ||
Yes | 42% | 1.74 | 1.38 | 2.21 | ||
ECOG PS (four levels) | ||||||
0 | 22% | Ref | 4.94E – 16 | 0.620 | ||
1 | 37% | 2.00 | 1.54 | 2.60 | ||
2 | 52% | 3.81 | 2.66 | 5.45 | ||
≥3 | 55% | 4.21 | 2.48 | 7.15 | ||
ECOG PS (two levels) | ||||||
0–1 | 29% | Ref | 7.06E – 12 | 0.568 | ||
≥2 | 53% | 2.74 | 2.06 | 3.66 | ||
LDH | ||||||
<ULN | 20% | Ref | 1.63E – 17 | 0.617 | ||
≥ULN | 42% | 2.92 | 2.26 | 3.77 | ||
Normalized LDH (LDH/ULN) Continuous, spline | 1.97E – 17 | 0.660 | ||||
Bulky disease | ||||||
<10 cm | 30% | Ref | 7.13E – 08 | 0.556 | ||
≥10 cm | 49% | 2.25 | 1.68 | 3.01 | ||
B symptoms | ||||||
Absent | 29% | Ref | 8.51E – 06 | 0.560 | ||
Present | 42% | 1.74 | 1.36 | 2.21 | ||
Composite histology | ||||||
No | 34% | Ref | 2.37E – 02 | 0.522 | ||
Yes | 25% | 0.64 | 0.42 | 0.95 | ||
ALC (109/L, two levels) | ||||||
>1 | 26% | Ref | 1.36E – 10 | 0.582 | ||
≤1 | 44% | 2.21 | 1.73 | 2.81 | ||
ALC (109/L, continuous) Spline | 6.15E – 09 | 0.608 | ||||
COO (Hans) | ||||||
GCB | 24% | 0.61 | 0.43 | 0.87 | 5.58E – 03 | 0.551 |
Non-GCB | 34% | Ref | ||||
IPI | ||||||
0 | 7% | Ref | 9.28E – 23 | 0.648 | ||
1 | 23% | 4.11 | 1.94 | 8.71 | ||
2 | 28% | 5.33 | 2.55 | 11.13 | ||
3 | 43% | 10.14 | 4.89 | 21.02 | ||
4 | 48% | 12.88 | 6.04 | 27.43 | ||
5 | 60% | 20.63 | 8.41 | 50.58 | ||
NCCN IPI | ||||||
0–1 | 12% | Ref | 5.52E – 23 | 0.642 | ||
2–3 | 25% | 2.58 | 1.54 | 4.30 | ||
4–5 | 40% | 5.03 | 3.04 | 8.34 | ||
6–8 | 60% | 11.38 | 6.36 | 20.38 | ||
0–8 scoring | 6.59E – 29 | 0.659 |
Note: Univariate models based on complete case data.
Multivariable modeling was performed using variables listed above. COO was excluded due to >50% of patients with missing data. The final model included age (continuous for >70 years), stage (four categories), ECOG PS (three levels), normalized LDH (continuous with inflection point at 1.3), bulky disease, ALC (continuous between 0 and 2 × 109 per L), and sex (Supporting Information Table 1); extranodal sites (number of sites or NCCN-IPI definition), B symptoms, and composite histology were not retained. The candidate model had excellent calibration (Supporting Information Fig. 2a) and good discrimination (c-statistic = 0.723, 95% CI: 0.699–0.755) which, as expected due to model overfitting, was higher than the independently developed IPI (c-statistic = 0.662, 95% CI: 0.32–0.69) and the NCCN-IPI (c-statistic = 0.677, 95% CI: 0.647–0.706) models.
Patient characteristics in the validation dataset
An independent dataset of 1,177 patients with pathology confirmed DLBCL was established for validation of the candidate model. Median age of patients in the dataset was slightly older than in the discovery dataset at 66 years (range 18–95) (Supporting Information Table 4). Similar to the discovery dataset, 54% of patients were male, most patients (72%) had stage III–IV disease, and 61% had an elevated LDH. At a median follow-up of 39 months (range 1–142), 529 patients (45%) had an event and 393 patients (33%) had died. The EFS24 failure rate was slightly higher in the validation dataset with 65% of patients achieving EFS24 (Kaplan-Meier estimate) and 35% failing to achieve EFS24.
Model performance
The multivariable model from the discovery dataset was applied to the independent validation dataset. The multivariable model retained good discrimination (c-statistic = 0.671, 95% CI: 0.642–0.702) in the validation dataset, which was superior to the IPI (c-statistic = 0.649, 95% CI: 0.616–0.680) and NCCN-IPI (c-statistic = 0.657, 95% CI: 0.626–0.687) (Table II). To provide the most accurate prognostic risk estimates for future application of the model in the clinical setting, we recalibrated the candidate model parameter estimates by refitting the multivariable model on the combined dataset of 2,525 patients [16]. The IPI24 (Fig. 1) is based on these parameter estimates (Table II).
TABLE II.
Final multivariable odds ratios for risk of failure to achieve EFS24 on full dataset (N = 2,525)
IPI24 | ||
---|---|---|
OR | 95% CI | |
Age | ||
Per year over 70 | 1.06 | 1.03–1.08 |
Stage | ||
I | Ref | |
II | 1.06 | 0.87–1.29 |
III | 1.16 | 0.96–1.40 |
IV | 1.50 | 1.30–1.73 |
ECOG PSa | ||
0 | Ref | |
1 | 1.00 | 0.88–1.13 |
≥2 | 1.39 | 1.19–1.62 |
Normalized LDHb | ||
Per unit increase if ≤1.3 | 2.57 | 1.67–4.05c |
Per unit increase if >1.3 | 1.28 | 1.16–1.42c |
Bulky disease | ||
<10 cm | Ref | |
≥10 cm | 1.28 | 1.02–1.60 |
ALC | ||
Per unit increase 109/L | 0.81 | 0.68–0.97 |
Sex | ||
Female | Ref | |
Male | 1.36 | 1.13–1.62 |
ECOG PS categories for 0 and 1 were collapsed in the implementation of the final model due to identical risk
Normalized LDH = LDH/ULN.
Bootstrap estimates from segmented model.
Figure 1.
IPI24 nomogram.
Model application and clinical utility
The IPI24 score estimates the probability of failing to achieve the EFS24 endpoint for an individual patient; in other words, the patient’s risk of an event or death prior to 24 months after diagnosis. The median predicted score was 0.36 (range 0.12–0.88) in the combined dataset, compared to the actual EFS24 failure rate of 35%, which represents general model calibration. Concordance for the IPI24 in the combined dataset was c-statistic = 0.708 (95% CI: 0.679–0.722). Approximately one in five patients (21%) had an IPI24 score >0.50, identifying a high risk subset of patients. These patients had an actual EFS24 failure rate of 58% in the combined dataset. We examined risk reclassification of the IPI when applying the IPI24 model in the full dataset. Within an IPI category there was a wide range of IPI24 scores with the IPI24 showing a risk gradient within the IPI categories; conversely the IPI does not add information in IPI24 groupings (Fig. 2). Generation of IPI24 scores in comparison to IPI classifications for sample patients are displayed in Supporting Information Table 2. In a sensitivity analysis, we rebuilt the model in patient subsets, including excluding patients with a concomitant low-grade lymphoma or patients receiving R-ACVBP on a clinical trial. Excluding these patients had little effect on individual patient risk prediction (see Supporting Information Materials).
Figure 2.
Distribution of IPI24 score by IPI classification in N = 2,525 patients. The values in red denote the mean EFS24 failure rate by group, where IPI24 score grouped as follows: (0–20%) vs. (20–30%) vs. (30–40%) vs. (40–50%) vs. (>50%).
Model implementation
To ease implementation of this model in the clinical setting, we have developed an e-version of the nomogram that can calculate the IPI24 risk score at the point of clinical encounter. An electronic version of the IPI24 algorithm can be found on the internet at QxMD (www.qxmd.com) and has been incorporated in the QxCalculate smartphone apps for iPhone and Android (http://www.qxmd.com/apps/calculate-by-qxmd). Parameter estimates for the IPI24 model can be found in the Supporting Information material. Additional details on the algorithm for incorporation into datasets for research or clinical purposes are available from the corresponding author.
Discussion
We present a novel, individualized prognostic calculator for newly diagnosed patients with DLBCL treated with immunochemotherapy. The calculator is the first model to use the EFS24 endpoint, a DLBCL-specific outcome which can be used in clinical trials or biologic studies of newly diagnosed DLBCL. The IPI24 estimates the individual probability of failing to achieve EFS24 using widely available clinical variables, more accurately reclassifies prognosis on a substantial number of patients over the standard IPI or NCCN-IPI, and identifies a significant percentage of patients with high risk disease, with over one in five patients having at least a 50% risk of failing to achieve EFS24 in our dataset.
While a number of studies have developed updates of the IPI in the rituximab era [17,18] including the recent NCCN-IPI [19], all of these models are predicated on using a time-to-event endpoint such as EFS or overall survival and restricts risk classification into broad categories (e.g., 0–5) rather than providing individual probabilities. Models using continuous progression-free survival (PFS) and/or OS include late events (after 24 months), a majority of which are not disease-related as we have previously shown [5]. The IPI and related indices are also highly influenced by age, since age is a strong predictor of continuous time-to-event outcomes such as overall survival in the general population. In contrast, an outcome evaluated at 2 years such as EFS24 is less influenced by age due to the 2-year timeframe and focus on disease-events. This allows us to generate patient-specific outcome prediction for a disease-specific endpoint that works well across a wide range of ages, further enhancing its utility.
The IPI24 contains the variables age, sex, serum LDH, Ann Arbor stage, bulky disease (>10 cm), ECOG PS, and ALC. These variables are all established clinical predictors of outcome in DLBCL. Age, LDH, stage, and ECOG PS remain from the original IPI with some modification. We included LDH on the normalized scale (LDH/upper limit of normal); the inclusion of LDH as a continuous variable resulted in LDH being the most influential variable in the model. Indeed, there was a large increase in risk with increased LDH even in patients with LDH in the normal range, suggesting cutoffs for LDH are underutilizing the prognostic information in this serum marker. Age was included as a continuous variable but only above 70 years, as age was not predictive in patients under 70. The age-70 cutoff was also previously identified by the e-IPI study [18]. In addition, an outcome evaluated at 2 years such as EFS24 is not as influenced by age as a continuous time-to-event outcomes such as OS, where age is predictive in the general population at large. However, age becomes more predictive of 2-year survival in elderly patients where the overall risk of dying in the next 2 years is clinically meaningful in the general population. ECOG PS retained its current grouping of 0–1 versus 2–4 when considered with the other variables in the model. Patients with bulky disease (>10 cm) had increased risk, consistent with previously published data [20]; the 10 cm dichotomization was the only cutoff available in our data, other cutpoints or a continuous scale may allow for additional prognostication. The prognostic significance of ALC has been shown by several studies in DLBCL [21]. Examination of the functional form showed a continuous prognostic range for ALC from 0 to 2. Incorporation of monocyte count in an ALC/AMC ratio may provide further information to the model [22]. There was an association of male sex with inferior EFS24, which although not prognostic alone, was significant after accounting for the other clinical factors. Its inclusion is supported by recent work suggesting differences in rituximab pharmacokinetics based on age and sex [23].
Our model is built using a large set of patients treated with standard of care for DLBCL, from both clinical trials and epidemiology registries. Primary model building was performed on all cases, including those with concomitant low-grade lymphoma as well as those treated on experimental arms of clinical trials. The diversity of the patient treatment and presentation reflects general practice and should allow for broad application of the model in the clinical setting. Sensitivity analyses showed that exclusion of these patients had little impact on the individual patient estimates of achieving EFS24. We also chose to restrict variables for potential inclusion in the model to standard clinical variables that routinely would be available to clinicians or researchers. This is a limitation of the model; however, the clinical variables in the model have strong history in regards to their clinical relevance in DLBCL and are thus very unlikely to be spurious associations. Assessment of tumor genetic features such as MYC translocations [24], tumor molecular subclassification [25], incorporation of host genetics [26], serum markers [6,7,27], or inclusion of other tumor biomarkers [28] will likely add predictive information but they are not yet routinely available in historical datasets in sufficient numbers to include in this analysis. Additional work will be needed in the future to incorporate these and other potential discoveries in a refined biologic model. The IPI24 establishes the clinical component for future predictive model building using the EFS24 endpoint in DLBCL.
COO per Hans algorithm [15] was available only on a subset of patients (n = 611) as many of these patients were diagnosed prior to standard clinical assessment of COO. COO as assessed by Hans algorithm showed very modest prognostic ability in our dataset (OR = 0.82, 95% CI: 0.54–1.24, P = 0.35) once the other clinical variables were included. There is mixed data in the literature on the prognostic significance of COO as assessed by Hans algorithm; association with the EFS24 endpoint has not previously been evaluated. As refinements in paraffin based assessment of molecular subtypes [25] gain clinical acceptance, the role of COO in the model can be reassessed.
Strengths of this study include the large number of patients (n = 2,525) from the US and Europe, as well as a mix of clinical trials and an epidemiology registry, which should provide excellent generalizability. We used robust modeling techniques and have generated the model from patients initially used to establish the EFS24 endpoint and validated the model in a large and diverse independent set of patients. Nevertheless, the diagnostic accuracy of the recalibrated IPI24 should be verified in additional external studies and randomized clinical trials. Future work will also need to evaluate the model in other racial/ethnic populations.
To facilitate clinical use, we have made the IPI24 available in paper form, on the web, or as an electronic app. The use of nomograms has become common in the medical community and allows creation of more elaborate models than simple scores as used by the current IPI. Most clinicians now routinely use online tools, tablets, and smartphones as part of their practice [29,30] and thus use of the IPI24 can be easily accomplished in the clinic and hospital settings. Risk estimates generated from the model can be used for patient counseling and risk stratification and can help inform treatment decisions. Future work will be needed to determine the optimal cutoff(s) of EFS24 risk regarding therapeutic decisions, such as identifying high risk patients for more aggressive induction therapy or consolidation with autologous stem cell transplantation.
In summary, we present a novel, improved clinical prognostic calculator for DLBCL in the immunochemotherapy era, using the new EFS24 endpoint. This model provides an individual risk prediction for a patient and can easily be calculated using online or smartphone based applications. The IPI24 is the first assessment of prognosis for EFS24 in DLBCL given our current set of standardly collected clinical variables.
Supplementary Material
Acknowledgments
The authors thank Robin Adams for editorial assistance.
Contract grant sponsors: National Institutes of Health to the University of Iowa/Mayo Clinic Lymphoma Specialized Program of Research Excellence; The Henry J. Predolin Foundation.
Footnotes
Additional Supporting Information may be found in the online version of this article.
References
- 1.Coiffier B, Lepage E, Briere J, et al. CHOP chemotherapy plus rituximab compared with CHOP alone in elderly patients with diffuse large-B-cell lymphoma. N Engl J Med. 2002;346:235–242. doi: 10.1056/NEJMoa011795. [DOI] [PubMed] [Google Scholar]
- 2.Habermann TM, Weller EA, Morrison VA, et al. Rituximab-CHOP versus CHOP alone or with maintenance rituximab in older patients with diffuse large B-cell lymphoma. J Clin Oncol. 2006;24:3121–3127. doi: 10.1200/JCO.2005.05.1003. [DOI] [PubMed] [Google Scholar]
- 3.Pfreundschuh M, Trumper 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;7:379–391. doi: 10.1016/S1470-2045(06)70664-7. [DOI] [PubMed] [Google Scholar]
- 4.Friedberg JW. Relapsed/refractory diffuse large B-cell lymphoma. Hematol Am Soc Hematol Educ Program. 2011;2011:498–505. doi: 10.1182/asheducation-2011.1.498. [DOI] [PubMed] [Google Scholar]
- 5.Maurer MJ, Ghesquieres H, Jais JP, et al. Event-free survival at 24 months is a robust end point for disease-related outcome in diffuse large B-cell lymphoma treated with immunochemotherapy. J Clin Oncol. 2014;32:1066–1073. doi: 10.1200/JCO.2013.51.5866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Drake MT, Maurer MJ, Link BK, et al. Vitamin D insufficiency and prognosis in non-Hodgkin’s lymphoma. J Clin Oncol. 2010;28:4191–4198. doi: 10.1200/JCO.2010.28.6674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Maurer MJ, Micallef IN, Cerhan JR, et al. Elevated serum free light chains are associated with event-free and overall survival in two independent cohorts of patients with diffuse large B-cell lymphoma. J Clin Oncol. 2011;29:1620–1626. doi: 10.1200/JCO.2010.29.4413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Nowakowski GS, Maurer MJ, Habermann TM, et al. Statin use and prognosis in patients with diffuse large B-cell lymphoma and follicular lymphoma in the rituximab era. J Clin Oncol. 2010;28:412–417. doi: 10.1200/JCO.2009.23.4245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Micallef IN, Maurer MJ, Wiseman GA, et al. Epratuzumab with rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone chemotherapy in patients with previously untreated diffuse large B-cell lymphoma. Blood. 2011;118:4053–4061. doi: 10.1182/blood-2011-02-336990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Recher C, Coiffier B, Haioun C, et al. Intensified chemotherapy with ACVBP plus rituximab versus standard CHOP plus rituximab for the treatment of diffuse large B-cell lymphoma (LNH03-2B): An open-label randomised phase 3 trial. Lancet. 2011;378:1858–1867. doi: 10.1016/S0140-6736(11)61040-4. [DOI] [PubMed] [Google Scholar]
- 11.Fitoussi O, Belhadj K, Mounier N, et al. Survival impact of rituximab combined with ACVBP and upfront consolidation autotransplantation in high-risk diffuse large B-cell lymphoma for GELA. Haematologica. 2011;96:1136–1143. doi: 10.3324/haematol.2010.038109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Peyrade F, Jardin F, Thieblemont C, et al. Attenuated immunochemotherapy regimen (R-miniCHOP) in elderly patients older than 80 years with diffuse large B-cell lymphoma: A multicentre, single-arm, phase 2 trial. Lancet Oncol. 2011;12:460–468. doi: 10.1016/S1470-2045(11)70069-9. [DOI] [PubMed] [Google Scholar]
- 13.Ketterer N, Coiffier B, Thieblemont C, et al. Phase III study of ACVBP versus ACVBP plus rituximab for patients with localized low-risk diffuse large B-cell lymphoma (LNH03-1B) Ann Oncol. 2013;24:1032–1037. doi: 10.1093/annonc/mds600. [DOI] [PubMed] [Google Scholar]
- 14.Delarue R, Tilly H, Mounier N, et al. Dosedense rituximab-CHOP compared with standard rituximab-CHOP in elderly patients with diffuse large B-cell lymphoma (the LNH03-6B study): A randomised phase 3 trial. Lancet Oncol. 2013;14:525–533. doi: 10.1016/S1470-2045(13)70122-0. [DOI] [PubMed] [Google Scholar]
- 15.Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103:275–282. doi: 10.1182/blood-2003-05-1545. [DOI] [PubMed] [Google Scholar]
- 16.Steyerberg EW, Borsboom GJ, van Houwelingen HC, et al. Validation and updating of predictive logistic regression models: A study on sample size and shrinkage. Stat Med. 2004;23:2567–2586. doi: 10.1002/sim.1844. [DOI] [PubMed] [Google Scholar]
- 17.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;109:1857–1861. doi: 10.1182/blood-2006-08-038257. [DOI] [PubMed] [Google Scholar]
- 18.Advani RH, Chen H, Habermann TM, et al. Comparison of conventional prognostic indices in patients older than 60 years with diffuse large B-cell lymphoma treated with R-CHOP in the US Intergroup Study (ECOG 4494, CALGB 9793): Consideration of age greater than 70 years in an elderly prognostic index (E-IPI) Br J Haematol. 2010;151:143–151. doi: 10.1111/j.1365-2141.2010.08331.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.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;123:837–842. doi: 10.1182/blood-2013-09-524108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Pfreundschuh M, Ho AD, Cavallin-Stahl E, et al. Prognostic significance of maximum tumour (bulk) diameter in young patients with good-prognosis diffuse large-B-cell lymphoma treated with CHOP-like chemotherapy with or without rituximab: An exploratory analysis of the MabThera International Trial Group (MInT) study. Lancet Oncol. 2008;9:435–444. doi: 10.1016/S1470-2045(08)70078-0. [DOI] [PubMed] [Google Scholar]
- 21.Kim DH, Baek JH, Chae YS, et al. Absolute lymphocyte counts predicts response to chemotherapy and survival in diffuse large B-cell lymphoma. Leukemia. 2007;21:2227–2230. doi: 10.1038/sj.leu.2404780. [DOI] [PubMed] [Google Scholar]
- 22.Porrata LF, Ristow K, Habermann T, et al. Predicting survival for diffuse large B-cell lymphoma patients using baseline neutrophil/lymphocyte ratio. Am J Hematol. 2010;85:896–899. doi: 10.1002/ajh.21849. [DOI] [PubMed] [Google Scholar]
- 23.Pfreundschuh M, Muller C, Zeynalova S, et al. Suboptimal dosing of rituximab in male and female patients with DLBCL. Blood. 2014;123:640–646. doi: 10.1182/blood-2013-07-517037. [DOI] [PubMed] [Google Scholar]
- 24.Savage KJ, Johnson NA, Ben-Neriah S, et al. MYC gene rearrangements are associated with a poor prognosis in diffuse large B-cell lymphoma patients treated with R-CHOP chemotherapy. Blood. 2009;114:3533–3537. doi: 10.1182/blood-2009-05-220095. [DOI] [PubMed] [Google Scholar]
- 25.Scott DW, Wright GW, Williams PM, et al. Determining cell-of-origin subtypes of diffuse large B-cell lymphoma using gene expression in formalin-fixed paraffin-embedded tissue. Blood. 2014;123:1214–1217. doi: 10.1182/blood-2013-11-536433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Habermann TM, Wang SS, Maurer MJ, et al. Host immune gene polymorphisms in combination with clinical and demographic factors predict late survival in diffuse large B-cell lymphoma patients in the pre-rituximab era. Blood. 2008;112:2694–2702. doi: 10.1182/blood-2007-09-111658. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Ansell SM, Maurer MJ, Ziesmer SC, et al. Elevated pretreatment serum levels of interferon-inducible protein-10 (CXCL10) predict disease relapse and prognosis in diffuse large B-cell lymphoma patients. Am J Hematol. 2012;87:865–869. doi: 10.1002/ajh.23259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Johnson NA, Slack GW, Savage KJ, et al. Concurrent expression of MYC and BCL2 in diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30:3452–3459. doi: 10.1200/JCO.2011.41.0985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Franko OI, Tirrell TF. Smartphone app use among medical providers in ACGME training programs. J Med Syst. 2012;36:3135–3139. doi: 10.1007/s10916-011-9798-7. [DOI] [PubMed] [Google Scholar]
- 30.Payne KB, Wharrad H, Watts K. Smartphone and medical related App use among medical students and junior doctors in the United Kingdom (UK): A regional survey. BMC Med Inform Decis Mak. 2012;12:121. doi: 10.1186/1472-6947-12-121. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.