Stand-alone intrathecal CNS prophylaxis provide unclear benefit in reducing CNS relapse risk in elderly DLBCL patients treated with R-CHOP and is associated increased infection-related toxicity

DLBCL CNS relapse following R-CHOP occurs in 2-5%. Many patients ≥70 years are unsuitable for high-dose methotrexate (HDMTX) prophylaxis and therefore often receive stand-alone intrathecal prophylaxis. The CNS international prognostic index (CNS-IPI) is a clinical CNS relapse risk score which has not specifically been validated in elderly patients. The value of CNS prophylaxis in patients ≥70 years remains uncertain. Data on 690 consecutively R-CHOP-treated DLBCL patients ≥70 years were collected across 8 UK centres (2009-2018). CNS prophylaxis was administered per physician preference. Median age was 77.2 years and median follow-up 2.8 years. CNS-IPI was 1-3 in 60.1%, 4 in 23.8%, 5 in 13.0% and 6 in 3.3%. Renal and/or adrenal (R/A) involvement occurred in 8.8%. 2-year CNS relapse incidence was 2.6% and according to CNS-IPI 1-3:0.8%, 4:3.6%, 5:3.8% and 6:21.8%. 2-year CNS relapse incidence for R/A was 10.0%. When excluding HDMTX (n=31) patients, there remained no change in unadjusted/adjusted CNS relapse for intrathecal prophylaxis effect according to CNS-IPI. CNS-IPI is valid in elderly R-CHOP-treated DLBCL patients with the highest risk in CNS-IPI 6 and R/A involvement. We observed no clear benefit for stand-alone intrathecal prophylaxis but observed an independent increased risk of infection-related admission during R-CHOP when intrathecal prophylaxis is administered. Introduction Diffuse large B cell lymphoma (DLBCL) is the commonest global lymphoid malignancy and is curable in 50-90% of patients with anthracycline-based immunochemotherapy depending on age, stage, patient fitness and disease biology (Davies, 2017). Systemic disease progression is the primary cause of treatment failure in DLBCL, however central nervous system (CNS) relapse occurs in approximately 2-5%, with very short median survivals (Schmitz et al, 20016; El-Galaly et al, 2017; Cabannes-Hamy et al, 2018; Tai et al, 2011; Boehme et al, 2009). Finding effective and non-toxic preventative strategies of CNS relapse is of critical importance across all age groups including the elderly. Several clinical and biological risk factors for CNS relapse have been described and include double hit (MYC plus BCL2 and/or BCL6 rearrangement) lymphoma (Oki et al, 2014), advanced stage III-IV disease, raised lactate dehydrogenase (LDH), >1 extranodal disease site and involvement of specific sites, for example, the testis, kidney or adrenal glands. The CNS-IPI (international prognostic index) used the standard IPI with an additional point for renal/adrenal (R/A) involvement to define the risk of CNS relapse in a large cohort of patients with aggressive B cell lymphoma (n=2164; 18-80 years; 80% DLBCL) (Schmitz et al, 20016). This tool classifies patients according to their baseline CNS-IPI score into risk categories. Low-risk (CNS-IPI 0-1; 46%), intermediate-risk (CNS-IPI 2-3; 41%), and high-risk (CNS-IPI 4-6; 12%) patients demonstrated a cumulative 2-year risk of CNS relapse of 0.6% (95% confidence interval (CI) 0-1.2%), 3.4% (95% CI 2.2-4.4%), and 10.2% (95% CI 6.3-14.1%) respectively, with similar results seen within the British Columbia Cancer Agency validation data set. A separate validation study (El-Galaly et al, 2017) by El-Galaly and colleagues analysed 1532 patients with a median age of 65 years who were staged with PET-CT and treated with R-CHOP/R-CHOP-like regimens. Four percent of patients developed CNS relapse and the multivariable analysis found that baseline stage III/IV disease, raised LDH, kidney/adrenal and uterine/testicular DLBCL involvement were independent risk factors for CNS relapse. Patients with >2 extranodal sites had a 3-year cumulative CNS relapse incidence of 15.2% compared to 2.6% in patients with ≤2 extranodal sites (p<0.001). Predictive tools like these have enabled targeted CNS prophylaxis in higher risk patients. Although the CNS-IPI is validated in younger (18-80 years) DLBCL patients (Schmitz et al, 20016) very few studies have assessed the incidence of CNS relapse, its associated risk factors, or the influence of CNS prophylaxis in elderly DLBCL patients treated with curative intent with R-CHOP (full or attenuated dose). A subgroup analysis from El-Galaly et al suggested that the CNS-IPI predicts CNS relapse in 494 patients ≥70 years (personal communication) (El-Galaly et al, 2017). There are few data pertaining directly to the elderly. However, a recent pooled analysis of two LYSA attenuated CHOP plus anti-CD20 phase II trials examined 270 trial-fit, selected patients ≥80 years (Cabannes-Hamy et al, 2018). In this analysis, no patients received CNS prophylaxis, and the CNS relapse risk was 1.8% at 2 years and 3% overall. The CNS relapse risk did not differ according to CNS-IPI (low-intermediate 3% (versus (vs.) high 2.8%; p=1.00). The majority of CNS relapse events in the rituximab era are intraparenchymal (Guirguis et al, 2012; Mitrovic et al, 2012), leading to an ongoing debate about the relative benefits of intrathecal (IT) prophylaxis and the more widespread adoption of high dose intravenous (IV) methotrexate (HDMTX) prophylaxis in younger (often <70 years) patients with adequate renal function and high risk clinical features (Savage, 2017). Although a small number of selected patients ≥70 years receive IV HDMTX and / or IT MTX, there is a relative lack of definitive evidence documenting the benefit of this approach and a lack of toxicity data utilising HDMTX in patients ≥70 years. Patients ≥70 years represent the minority of those studied in post hoc analyses from the RICOVER-60, RCHOP 14 versus 21 trials (for example, 216/1080 (20%)) (Gleeson et al, 2017; Boehme et al, 2009), GOYA (Klanova et al, 2019) and retrospective series (Tomita et al, 2015; Kumar et al, 2012; El-Galaly et al, 2017). Collectively, these series failed to demonstrate any consistent benefit of IT prophylaxis across a broad range of ages. Furthermore, the relative benefit of CNS prophylaxis is uncertain in older patients who are more vulnerable to its potential toxicities (Savage, 2017) and this question has not been addressed. We conducted a retrospective multicentre national analysis of the risk of CNS relapse in patients ≥70 years old with DLBCL treated with curative intent. We aimed to further validate the CNS-IPI in an elderly population, assess the influence of CNS prophylaxis on CNS relapse risk, and analyse its toxicity patterns in elderly patients. Methodology Data on 690 consecutively treated first line elderly DLBCL patients of ≥70 years were retrospectively collected across 8 UK centres from 2007-2018. All patients were untreated, de novo DLBCL or untreated transformed (to DLBCL) indolent B cell non-Hodgkin lymphoma (iNHL). Patients with leg-type DLBCL, post transplantation lymphoproliferative disease, concurrent CNS involvement, HIV-positivity and pre-treated transformed iNHL were excluded. All patients were treated with between 1-8 cycles of full or attenuated dose R-CHOP with curative intent. Baseline CNS evaluation was not mandated for inclusion in the study although this was performed routinely in patients with clinically concerning features of CNS involvement. We collected baseline disease characteristics including the CNS-IPI and extranodal site involvement. Multiple lesions within one organ or tissue type (e.g. multiple skin or pulmonary lesions) were considered as a single extranodal site. Bone involvement was defined by focal lesions on PET-CT or standard CT, and/or by bone marrow involvement at biopsy. CNS prophylaxis was administered as per local protocol according to physician preference. Admission reason(s) and rate were collected following IV HDMTX. Patients were analysed according to whether they had received no prophylaxis, IT prophylaxis only, HDMTX only, or a combination of IT MTX and IV HDMTX. During the inclusion period, treating physicians typically administered IT MTX synchronously with R-CHOP and typically proceeded with up to 2 cycles of HDMTX following R-CHOP in selected patients. CNS relapse was defined by intraparenchymal, spinal cord, leptomeningeal or ocular recurrence which was documented either by histopathological, cytological or clinico-radiological features. Statistical analysis Baseline patient characteristics were summarised in a descriptive manner. Survival analyses were performed using Kaplan-Meier survival analysis (Kaplan & Meier, 1958) and Cox regression with comparisons between categories were made using the log-rank test. All time to event analyses were measured from the date of the initial DLBCL diagnosis until the event with patients censored at the date last seen if alive and event free. Time to CNS relapse and the cumulative incidence of CNS relapse at 2 and 3 years were analysed using competing risk survival analyses with death and systemic relapse counted as competing events. Survival following the time of relapse was measured as the time from relapse to either the date of death from any cause or censored at the last clinical contact. Patient follow-up was censored in February 2019. Univariable analyses of potential influencing factors for CNS relapse including the CNS-IPI, specific extranodal sites and use of CNS prophylaxis were performed. Multivariable analysis assessment the risk of infective re-admission was assessed by logistic regression analysis. For this analysis, the specific time point from which patients were considered at risk of infection was from the start of cycle 1 and during R-CHOP immunochemotherapy i.e. until 3 weeks following the final cycle. Infective admission for each patient was collected in a binary fashion. Analyses of admission data and the risk of CNS relapse according to CNS prophylaxis have been repeated for only patients who were alive and progression free at 6 months to attempt to resolve possible issues with immortality bias. Statistical analyses were performed in Stata 15.1 (StataCorp, College Station, TX, USA). 95% confidence intervals are presented