Outcomes of Primary Central Nervous System Lymphoma (PCNSL) Treated with Immunochemotherapy or Chemotherapy Alone: Analysis of the National Cancer Data Base 2013-2018

Background: PCNSL is an aggressive lymphoma typically treated with high-dose methotrexate-based chemotherapy. Rituximab, a CD20-directed immunotherapy, is frequently incorporated into the treatment plan, although in randomized trials it did not improve survival [Bromberg et al., Lancet Oncol 2019; Ferreri et al., Lancet Haematol 2016]. Since 2013, the National Cancer Data Base (NCDB), which contains records of >80% lymphomas diagnosed in the United States [US]), has distinguished "chemotherapy" and "immunotherapy" as separate components of initial treatment for PCNSL. We examined factors associated with receipt of immunotherapy (or chemotherapy alone) and resulting outcomes. Methods: Using the NCDB data from 2013-2018, we identified patients with PCNSL (confirmed large B-cell histology) and excluded those who did not receive chemotherapy within 60 days of diagnosis, or those who received radiotherapy within 30 days of diagnosis. Patients were grouped according to type of first-line regimen as receiving chemotherapy alone or immunochemotherapy (defined as immunotherapy starting no later than 30 days from the first chemotherapy). We studied factors associated with the receipt of immunochemotherapy in a multivariable regression for relative risk (RR), with random intercept to account for hospital-specific treatment selection process. We then matched patients who received chemotherapy alone or immunochemotherapy using 1:1 propensity score matching to balance all available confounders between the groups. Overall survival (OS) was compared in the matched cohort. All estimates include 95% confidence intervals (CI). Results: We identified 4,691 eligible patients with PCNSL (median age, 66 years [y]; 49% women; 3% HIV+). Median OS in the entire cohort was 4.1 y (95% CI, 3.9-4.6). Concurrent immunotherapy was administered to 3,147 (67%) patients, and this proportion increased from 45% in 2013 to 76% in 2018 (Fig. A). In a multivariable model, the main factors associated with the use of immunotherapy included Black race (RR=0.88; 95%CI: 0.79-0.99 - compared with White non-Hispanic [WNH]) or Hispanic ethnicity (RR=0.88; 95%CI: 0.80-0.97 - compared with WNH), HIV+ status (RR=0.79; 95%CI: 0.66-0.93), male sex (RR=0.95; 95%CI: 0.91-0.99), treatment in a lower-volume hospital (RR=1.16 for hospitals with >5 PCNSL cases/y; 95%CI: 1.07-1.27), and year of diagnosis. There were no significant differences by age, number of comorbidities, type of chemotherapy (single vs. multi-agent), anatomical location of PCNSL (brain, spinal cord, leptomeningeal disease), or type of hospital (academic vs. community). We found evidence for a strong preference for use/non-use of immunotherapy in each hospital (P<.0001 against a model without the random intercept). We matched 2,784 patients in a 1:1 ratio according to the receipt of immunochemotherapy or chemotherapy alone, successfully minimizing any between-group differences in all observed confounders. In the matched cohort, median OS for recipients of immunochemotherapy was 4.6 y (95%CI, 3.9-5.5) compared with 2.8 y (95% CI, 2.3-3.2) for recipients of chemotherapy alone; 2-years OS estimates were, respectively, 62% (95%CI, 59-65%) and 55% (95%CI, 52-57%). Receipt of immunotherapy was associated with significantly better OS (hazard ratio 0.80, 95%CI, 0.73-0.89; Fig. B). We observed no heterogeneity by propensity score quintile or type of cytotoxic chemotherapy (single or multi-agent). However, there was heterogeneity by age, as the advantage of immunotherapy was only observed for patients aged ≤75 y (HR=0.76; 95%CI: 0.67-0.85), but not for those older than 75 y (HR=0.94; 95%CI: 0.76-1.16). Conclusions: In this large real-world dataset summarizing outcomes of most patients treated in the US for PCNSL in 2013-2018, the use of immunotherapy (presumably, mainly rituximab) in addition to chemotherapy has increased over time, despite uncertain benefit. The use of immunochemotherapy was associated with sociodemographic variables and local (hospital-level) preference rather than clinical factors (except for the rare HIV status). After accounting for confounding factors, OS was better for patients receiving immunotherapy who were younger than 75y at diagnosis. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal