Background: Aggressive non-Hodgkin lymphomas with secondary central nervous system (CNS) involvement bear a dismal prognosis. Optimal treatment remains so far unclear, and effective treatment options remain an unmet clinical need. Remission rates are in general low, resulting in rapid relapses and palliative care in the majority of patients. High-intensity treatment combining effective CNS-directed chemoimmunotherapy with autologous stem cell transplantation was shown in a recent phase 2 trial to induce durable remissions. Here, we report the outcome of the first real-world patient cohort treated according to the published protocol. Methods: We retrospectively identified 17 HIV-negative lymphoma patients with secondary CNS involvement, either at first diagnosis or at relapse of lymphoma, treated according to the study protocol published by Ferreri et al. [J Clin Oncol. 2015] at two university medical centers in Germany. Treatment consisted of four cycles of chemoimmunotherapy with a consolidating autologous stem cell transplantation. Adverse events and overall outcome were assessed. Results: Five patients had CNS involvement at first diagnosis and 12 patients at relapse of lymphoma. A complete response was achieved in 9 patients. Median survival was 11 months. Five patients died of septic complications and 4 patients succumbed to progression or relapse of disease. Conclusions: The outcome of our real-world cohort emphasizes the possible toxic character of the treatment protocol by Ferreri et al. [J Clin Oncol. 2015]. Further improvement in treatment regimens is still an unmet need.
Central nervous system (CNS) involvement of aggressive systemic non-Hodgkin’s lymphoma (SCNSL) at diagnosis or relapse of disease has a dismal prognosis [1, 2]. Around 4% of patients experience CNS involvement at initial diagnosis  and 1–10% of patients in the relapsed setting depending on a variety of risk factors . Involved sites include brain parenchyma, spinal cord, cranial nerves, eyes, or meninges, often resulting in substantial disabilities and significant impairment in quality of life [1, 4]. Reports of survival vary, with a published median overall survival ranging from 3.9 months  to 1.5 years .
Different treatment regimens in SCNSL have been published, incorporating substances known to cross the blood-brain barrier like procarbazine, etoposide, ifosfamide, thiotepa ,and carmustine (BCNU) in combination with methotrexate (MTX) and cytarabine followed by autologous stem cell transplantation (ASCT) if feasible [7, 8]. However, the number of clinical trials conducted in this distinct patient population is limited, since patients with CNS involvement are frequently excluded from clinical trials.
The first prospective phase 2 study conducted by Kor-fel et al.  included 30 patients and demonstrated the curative potential of combining CNS-directed substances with ASCT in patients with CNS involvement at relapse. With a toxicity-related death rate of only 3% and reduced CNS toxicity due to omitting whole brain irradiation, a remission rate of nearly 50% has been achieved, which is comparable to remission rates in patients without CNS involvement at relapse .
Building on this achievement, Ferreri et al.  conducted a multicenter phase 2 study including 38 patients with CNS involvement at relapse as well as CNS involvement at first diagnosis of lymphoma. By adding rituximab and omitting non-CNS-directed substances to reduce systemic toxicity, a 5-year overall survival of 68% for patients completing the full protocol (including ASCT) and 41% for the whole cohort was documented. With patients up to the age of 70 years, ECOG performance scores up to 3, and different lymphoma etiologies (diffuse large B-cell lymphoma [DLBCL] and follicular and blastoid mantle cell lymphoma), the study by Ferreri et al.  received acknowledgment in the community since it did not select for a small subset of patients that is overly fit and more easy to treat. Yet, the rather high toxicity-related death rate of 11% as well as the potential of undertreating patients at first diagnosis of lymphoma by excluding full doses of essential substances such as anthracyclines was pointed out [11, 12].
More recently, the results of the multicenter phase 2 MARIETTA trial by Ferreri et al.  were published. This trial investigates a similar regimen including chemoimmunotherapy based on MTX and ifosfamide followed by consolidation with high-dose chemotherapy and ASCT. With the inclusion of ifosfamide, they addressed concerns about the efficacy of cyclophosphamide, utilized in their previous study, to cross the blood-brain barrier. Although 2-year overall survival with 46% was not improved in comparison to previous studies, Ferreri et al.  demonstrated a better survival for patients with CNS involvement at first diagnosis, with 71% alive after 2 years and an improvement in the overall safety of the new protocol. In order to add real-world evidence to the advances described , we report the outcome of 17 patients treated according to the published treatment protocol at two academic centers in Germany.
In this retrospective analysis, we assessed the outcome of HIV-negative lymphoma patients with secondary CNS involvement treated according to the study protocol published by Ferreri et al. . Between 2016 and 2019, we identified 15 patients who were treated at the University Medical Center Hamburg-Eppendorf and 2 patients treated at the University Medical Center Halle (Saale). Patient characteristics are shown in Table 1. All patients experienced secondary CNS involvement either at primary diagnosis of systemic lymphoma or at time of lymphoma relapse and had no concomitant medical condition prohibiting high-dose chemotherapy. Treatment according to Ferreri et al.  was recommended by each institution’s Interdisciplinary Lymphoma Board for all 17 patients. All patients consented to the treatment recommendation and were thoroughly informed about the potential risks of the proposed intensive treatment regimen including alternative treatment plans.
Treatment consisted of two cycles of induction chemotherapy containing high-dose MTX, high-dose cytarabine, rituximab, and intrathecal (i.th.) therapy with dexamethasone, cytarabine, and MTX, followed by two cycles of an intensification phase containing high-dose cyclophosphamide, high-dose cytarabine, rituximab, and i.th. therapy. Patients with systemic residual disease received an additional cycle of high-dose etoposide, rituximab, and i.th. therapy. Finally, a consolidating ASCT with BCNU and thiotepa conditioning was performed if a remission was achieved. A treatment flowchart is displayed in Figure 1. For detailed treatment schedule and substance dosing, we refer to the published protocol by Ferreri et al. .
All patients received prophylactic anti-infective therapy and granulocyte colony-stimulating factor -containing agents after each cycle of chemotherapy in concordance with published ASCO recommendations [14, 15]. Monitoring for complications during chemotherapy-induced neutropenia was done on an inpatient basis.
Adverse effects equal or higher grade 3 were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) v3.0 guidelines. Response assessment after induction, intensification, and consolidation therapy was performed utilizing magnetic resonance imaging and computed tomography depending on the sites of lymphoma manifestation and was based on the National Cancer Institute standardized response criteria for lymphoma . Cerebrospinal fluid (CSF) was monitored by cytology and flow cytometry in patients with CSF manifestation at diagnosis of SCNSL.
Median follow-up time was calculated with a reverse Kaplan-Meier estimation (median time to censoring). Median survival time and 1-year overall survival of the Ferreri et al.  patient cohort were estimated from the reported Kaplan-Meier plot (see online suppl. Fig. 1; for all online suppl. material, see www.karger.com/doi/10.1159/000517531). Statistical calculations and plotting was performed with R , the survminer , and ggplot2  package.
Seventeen lymphoma patients with secondary CNS involvement at diagnosis or relapse, who received treatment according to the published protocol by Ferreri et al.  between 2016 and 2019 at two academic centers in Germany, were retrospectively identified. The majority of patients presented with DLBCL (n = 14), 5 of which had newly diagnosed DLBCL with CNS involvement and 9 had relapsed disease. One patient presented with a relapsed post-kidney transplantation DLBCL, 1 patient with an aggressively relapsed Waldenstrom macroglobulinemia, and 1 patient with an aggressively transformed marginal zone lymphoma. Aggressive systemic and CNS-directed treatment was indicated in all 3 patients. Median age for our cohort was 52 years (range 34–71 years); 8 patients had an elevated LDH level and 5 patients had an ECOG performance status of 2 or higher (see Table 1). Systemic and CNS sites of lymphoma manifestation as well as bone marrow and CSF involvement are summarized in online suppl. Table 2. Five out of 12 patients with CNS involvement at relapse also had systemic lymphoma manifestations.
Previous treatment for the relapsed DLBCL patients, including the patient with post-kidney transplantation DLBCL, had been rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisolone (R-CHOP) (n = 10). Two patients had received rituximab and bendamustine as first-line treatment for a diagnosis of indolent lymphoma. None of the relapsed DLBCL patients in our cohort received prior prophylactic CNS-directed therapy.
Events and Outcome
Overall, 12 of 17 patients (71%) completed the treatment protocol (see Fig. 1). During treatment, 2 patients died of sepsis and 1 patient experienced progressive disease and was switched to best supportive care. Additionally, 2 patients had to be taken off protocol due to infectious complications warranting surgical intervention: 1 patient developed septic arthritis after the second cycle of induction therapy and died shortly after due to progressive disease and 1 patient developed aspergilloma and failed to mobilize stem cells after the first cycle of intensification chemotherapy. The latter patient has stayed in complete remission up to now after two additional cycles of rituximab, dexamethason, etoposide, ifosfamide, and carboplatin (R-DeVIC) (see Fig. 2).
Observed therapy-related toxicities were mostly myelotoxicity, which was manageable with transfusion, anti-infective, and granulocyte colony-stimulating factor-containing agent therapy. We did not observe any neurotoxicity grade 3 or higher. Other organ-related toxicity was minimal and non-severe (see online suppl. Table 1).
The overall response rate of our patient cohort was 59% (10 of 17), with 9 patients achieving a complete remission (53%) and 1 patient a partial remission with a stable CNS disease and a systemic complete remission. Estimated median follow-up time of our cohort was 22 months (interquartile range 13–46 months). Lethal events occurred in 9 patients (53%). Progressive and relapsing disease (n = 4, 24%) accounted for nearly half of the mortality in our cohort. Surprisingly, the other half of lethal events was caused by septic complications (n = 5, 29%), 3 of which occurred after the consolidating ASCT (see Fig. 2). All 3 patients were in complete remission before consolidation therapy. Four of the 5 fatal septic events occurred in pretreated patients with relapsed disease, 2 of which had a systemic and CNS relapse of lymphoma. One patient experienced a systemic and CNS lymphoma relapse 40 months after completion of the treatment protocol. He received a salvage therapy with two cycles of R-MTX-cytarabine in preparation for CD19-targeted CAR T-cell therapy.
Optimal treatment of lymphoma patients with concomitant CNS involvement at initial diagnosis or secondary involvement at relapse remains a difficult challenge. High-intensity chemotherapy followed by ASCT consolidation for eligible relapsed patients is able to induce long-term remissions with overall 2-year survival rates between 41 and 63% [9, 20]. Data on optimal treatment for patients with secondary CNS involvement at initial presentation are even scarcer [10, 21, 22]. Available retrospective analysis underlines the role of intensive CNS-directed chemotherapy for controlling CNS manifestation of lymphoma in these patients [23, 24]. However, the value of consolidating ASCT in this patient cohort is not well defined. Damaj et al.  retrospectively reported a significant survival benefit for patients receiving a consolidative ASCT (19 of 60 patients receiving ASCT with a 3-year overall survival of 75% compared to 29%), whereas Wight et al.  could not observe a significant difference for consolidative ASCT in patients treated with intensive CNS-directed therapy when matched for induction outcomes (38 of 80 patients receiving intensive CNS induction therapy, of which 14 received ASCT with a nonsignificant 2-year overall survival of 66% compared to 56%).
Ferreri et al.  were the first to prospectively show that a combination of high-intensity chemoimmunotherapy and ASCT consolidation in patients with SCNSL at relapse as well as at initial diagnosis is feasible and effective. Their convincing data led to the change of practice at our two institutions. During a 3-year time span, we treated 17 patients according to the protocol by Ferreri et al. , which is a substantial number of patients, considering the rarity of secondary CNS involvement in lymphoma patients. Thereby, we can add clinically relevant real-world evidence to the published study data.
Overall, our analyzed patient cohort resembles the cohort of the Ferreri et al.  study in all major aspects, including heterogeneity of included lymphoma entities, age distribution, elevated LDH levels, systemic and CNS sites of lymphoma manifestation, and bone marrow and CSF involvement (see Table 1). With 65% (11 of 17) of patients completing the protocol in comparison to 53% (20 of 38) reported by Ferreri et al. , the complete treatment schedule is certainly feasible at specialized hematological centers. Most therapy-related toxicities grade 3 or higher, including neutropenia and neutropenic fever, were consistent with the previously reported numbers . The overall response rate of our patient cohort with 10 responding patients including 9 complete remissions is in line with the Ferreri et al.  study cohort (OR: 59 vs. 63%, complete remission: 53 vs. 63%) .
With the limitation of a shorter estimated median follow-up time of our cohort (22 vs. 48 months), reasonably assessable survival metrics like median survival and 2-year overall survival were comparable to those of the Ferreri et al.  cohort (11 vs. 15 months and 46 vs. 44%, respectively; see online suppl. Fig. 1). The 2-year overall survival of 46% for our cohort is also in line with previously published retrospective and prospective outcome data, including the recently published MARIETTA trial (46% , 35% , 21%  and 53% , and 63% ). Overall, this poses a rather improved outcome for this unfavorable patient cohort.
Fatal events occurred in 9 patients (53%), congruent with the event rate reported by Ferreri et al.  (53%). However, 5 patients (29%) died of septic complications, which resulted in a much higher toxicity-related death rate in comparison to 11% (4 of 38) reported by Ferreri et al.  and 3% (1 of 30) reported by Korfel et al. . Although the size, heterogeneity, and shorter follow-up of our patient cohort prohibit a detailed statistical analysis and comparison, we find evidence for the previously articulated concern of Korfel et al.  of an increased number of toxicity-related deaths.
We did not find any significant outcome difference between patients with relapsing disease and patients with CNS involvement at first diagnosis. Moreover, no outcome differences could be observed between patients with relapsed disease limited to the CNS and patients with CNS and systemic lymphoma recurrence.
Taken together, the outcome of our real-world cohort emphasizes, although effective and feasible, the toxic character of the treatment protocol by Ferreri et al. . The postulated role as a standard of care treatment for patients with SCNSL remains therefore debatable. The recently published MARIETTA trial protocol was not able to further improve survival. However, a reduction of fatal toxic events was achieved. Further improvement in treatment regimens with higher efficacy and less toxicity in combination with improved risk stratification continues to be an unmet need. Novel treatments like CD-19-directed CAR T-cell therapy  might serve as an attractive salvage therapy option for relapsing SCNSL patients.
Statement of Ethics
This study was conducted in accordance with the Helsinki Declaration. All patients gave their written consent to the treatment recommendation by our Interdisciplinary Lymphoma Board, to data collection, and to potential retrospective data analysis. Our study protocol was reviewed and the need for approval was waived by the Ethical Committee of the Ärztekammer Hamburg.
Conflict of Interest Statement
The authors declare no competing interests.
No funding for this study was received.
Idea and design of the study: S.J., B.T., and M.B.; supply of data and infrastructure: M.B., C.F., J.D., K.C.W., C.B., and M.W.; data collection, analysis, and interpretation: B.T., S.J., S.S., and M.B.; and drafting of the manuscript: B.T., S.J., M.B., M.W., S.S., C.F., J.D., K.C.W., and C.B.
Consent for Publication
All authors consent for publication of the manuscript.