Abstract
Primary central nervous system lymphoma (PCNSL) is a rare form of lymphoma with a poor prognosis. We used the LOC network database, including 365 patients from 14 voluntary centers. Inclusion criteria were age ≥ 60 years, immunocompetence, confirmed PCNSL diagnosis and treatment with curative intent with MPV-based regimen (Methotrexate, Procarbazine, Vincristine) between 2011 and 2021. We analyzed LOS from first day of chemotherapy to dismissal from hospital. Using the MAXStat method we found the LOS threshold of 15 days, to be one of most statistically significant and most convenient to separate two groups with different OS. The median age of the whole population was 71 years, median Karnofsky performance score (KPS) was 70%. The main differences between short and long LOS groups were median KPS at diagnosis (70% vs 50%) and deep structure involvement status (51% vs 67%). With a median follow up of 22.8 months, the median OS was respectively 62.2 and 20 months for short and long LOS. Median PFS was respectively 18 and 9.3 months. Multivariate analysis found in our cohort that age was associated with OS (P = 0.01) and LOS with OS (P = 0.005) and PFS (P = 0.014). KPS at diagnosis is a cofounder to LOS in the analysis. LOS remains associated with OS (P = 0.016) in a subgroup analysis of patients with baseline KPS < 70%. Altogether, the LOS > 15 days identifies a population of patients with poorer outcome in elderly PCNSL patients. Its underlying drivers require further investigation in larger and prospective cohorts.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00277-025-06654-x.
Keywords: PCNSL, Chemotherapy, Length of stay, Prognosis
Dear Editor,
Patients with primary central nervous system lymphoma (PCNSL) still have a dismal prognosis and represent an important therapeutic challenge. Even though molecular testing may be useful for diagnosis, [1] or prognostication, [2] high-dose intravenous methotrexate (HD MTX) still remains the cornerstone of PCNSL treatment. In France, the regimen used for patients over 60 years of age combines rituximab with HD MTX, procarbazine, vincristine, and cytarabine (R-MPV-A) [3, 4]. In two prognostic scores published in the literature, [5, 6] common prognostic factors include age and general condition at diagnosis as important covariables for disease control and survival.
Following the observation that patients with PCNSL have often a prolonged length of stay (LOS) in the hospital, we explored the clinical significance of LOS in PCNSL.
Hospitalization LOS can be associated with several factors. HD MTX toxicity has been reported to increase the LOS [7] and may be associated with a pejorative outcome [8]. Moreover, it is an important etiology of renal failure, which is also associated with a prolonged LOS, even when extrarenal epuration is not necessary [9]. R-MPV-A has been reported to be associated with a risk of grade ≥ 3 neutropenia of 68% with few febrile neutropenia, which can also increase the LOS, even with prophylactic use of granulocytic colony-stimulating factors.
We hypothesized that a longer LOS in the hospital at the beginning of their treatment, may be associated with worse outcomes. To address this, we extracted data from the national Lymphome Oculo-Cérébral (LOC) registry. Data absent from the registry were collected on site by accessing patient records. Study was approved by the ethic committee of Tours University Hospital, in accordance with the 1964 Helsinki Declaration and its later amendments.
Inclusion criteria were: patients above 60 years, diagnosed between January 1, 2011, and January 1, 2021, and treated with a R-MPV-A backbone regardless of initially-planned consolidation treatment. Histological or cytological documentation of PCNSL was necessary, and an imaging excluding other lymphoma localizations was required. Exclusion criteria were: missing data for the primary endpoint, a history of systemic lymphoma, immunosuppression-related lymphoma, isolated intra-ocular lymphoma, histology other than diffuse large B-cell lymphoma, upfront palliative care, in-hospital mortality, and patient refusal to the collection of their data.
LOS was defined from the first day of chemotherapy to the date of discharge. Response was assessed in accordance with the International PCNSL Collaborative Group criteria [10].
Overall survival (OS) and progression free survival (PFS) and estimates were calculated using Kaplan–Meier curves. The LOS threshold was determined using the maximally selected rank statistic method (MaxStat R package), which was applied to identify the optimal cutoff differentiating overall survival (OS) between groups. The two groups were compared using, when appropriate, the Wilcoxon rank sum test, the Pearson Chi2 test, or the Fisher’s exact test. Univariate analysis using the Kaplan Meier method with the Logrank test was performed for each variable. Multivariate analysis was performed with the Cox proportional hazards regression model. Variables that were significant in univariate analysis were considered for the model. A two-sided p-value < 0.05 was considered significant.
Of the 1271 patients treated with an intent-to treat MPV-based regimen identified in the LOC registry as of March 31st, 2022, 365 patients from 13 centres of the LOC network were retained for data analysis (Fig. 1A). The median LOS was 6 days (range 2–160 days). Two LOS thresholds were found to statistically significant (Fig. 1B): 15 days and 23 days. We performed a ROC analysis on LOS for survival (Supplemental Fig. 1). The 15-day threshold showed high specificity for OS and PFS (0.85 and 0.83) but low sensitivity (0.38 and 0.31), indicating that discharge within 15 days was more strongly linked to favorable outcomes than prolonged hospitalization to poor outcomes. As well, the 23-days threshold had a high specificity of 0.91 and a lower sensibility of 0.28 for OS, respectively 0.92 and 0.23 for PFS. For further analysis, the 15-day threshold was selected instead of 23 days to prevent excessive imbalance in group sample sizes.
Fig. 1.
Description of the population, and the breakdown in the LOS groups. A. study flowchart indicating the reasons for exclusion of patients resulting in the setup of the analysis set. The PCNSL LOC registry was interrogated for patients aged 60 years or above treated with a MPV-A regimen. Treating centres were contacted for an authorization to retrieve hospitalization data. A clinical research assistant was mandated to collect the hospitalization data in responding centres, and verified that inclusion criteria were still present. B Graphic representation of the MaxStat output indicating the significance of the difference of OS according to the threshold used to separate the cohort in two LOS groups. The 15-day threshold was preferred to the 23-day threshold because of practicality, higher specificity, and increased power for inter-group comparisons. C Graphic representation of the median times between symptoms, admission to the hospital, biopsy, treatment and discharge, according to the two LOS groups. D Overall survival of patients of the LOS > 15d and LOS ≤ 15d groups
Patient observations were thus separated in two groups: LOS > 15d (95 patients, median LOS 5.5 days), and LOS ≤ 15d (270 patients, median LOS 31 days), and their main characteristics are presented in Table 1. The median age at diagnosis was 71 years (range 60–88 years) and the median Karnofsky performance status (KPS) at diagnosis was 70% (range: 10–100%).
Table 1.
Patient characteristics and LOS
| Available data (%) | Overall N = 365 |
LOS ≤ 15 days N = 270 |
LOS > 15 days N = 95 |
P value | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Wilcoxon | OS | PFS | ||||||||
| Age, years (IQR) | 100 | 71 | (67–77) | 71 | (66–77) | 71 | (68–77) | 0.7 | 0.017 | 0.05 |
| KPS, % (IQR) | 100 | 70 | (50–80) | 70 | (60–80) | 50 | (40–65) | < 0.001 | 0.021 | 0.47 |
| MSKCC score 3, n (%) | 100 | 167 | (48) | 96 | (36) | 71 | (75) | < 0.001 | 0.0014 | 0.16 |
| LDH > N | 70 | 111 | (43) | 84 | (41) | 71 | (75) | 0.39 | 0.15 | 0.075 |
| Elevated CSF protein level, n (%) | 67 | 191 | (78) | 133 | (75) | 58 | (87) | 0.046 | 0.81 | 0.81 |
| Meningeal involvement, n (%) | 75 | 45 | (16) | 31 | (16) | 14 | (19) | 0.5 | 0.8 | 0.87 |
| Deep structure involvement, n (%) | 100 | 201 | (55) | 137 | (51) | 64 | (67) | 0.005 | 0.2 | 0.19 |
| Clinical presentation | 100 | |||||||||
| Motor deficit | 149 | (41) | 96 | (36) | 53 | (56) | < 0.001 | 0.42 | 0.46 | |
| Sensory deficit | 38 | (10) | 22 | (8) | 16 | (17) | 0.01 | 0.24 | 0.18 | |
| Walking disorder | 200 | (55) | 136 | (50) | 64 | (67) | 0.004 | 0.018 | 0.09 | |
| Epilepsy | 28 | (8) | 22 | (8) | 6 | (6) | 0.56 | 0.003 | < 0.001 | |
| Intracranial Hypertension | 69 | (19) | 49 | (18) | 20 | (21) | 0.53 | 0.78 | 0.67 | |
| Cognitive impairment | 236 | (65) | 159 | (59) | 77 | (81) | < 0.001 | 0.14 | 0.5 | |
| Behavior disorder | 54 | (15) | 38 | (14) | 16 | (17) | 0.5 | 0.007 | 0.2 | |
| Sight perturbation | 80 | (22) | 61 | (23) | 19 | (20) | 0.59 | 0.31 | 0.48 | |
| B symptoms | 37 | (10) | 26 | (10) | 11 | (12) | 0.58 | 0.55 | 0.8 | |
| Treatments received | 100 | |||||||||
| MPV-A | 365 | (100) | 270 | (100) | 95 | (100) | 1 | 1 | 1 | |
| Radiotherapy | 3 | (1) | 1 | (0.004) | 2 | (2) | 0.17 | 0.21 | 0.35 | |
| Corticosteroids | 272 | (75) | 206 | (77) | 66 | (69) | 0.15 | 0.75 | 0.49 | |
| Hypertonic serum | 1 | (0) | 1 | (0.004) | 0 | (0) | 1 | 0.31 | 0.71 | |
| Rituximab | 308 | (84) | 232 | (86) | 76 | (80) | 0.17 | 0.37 | 0.46 | |
| Delayed MTX elimination | 69 | (19) | 45 | (17) | 24 | (25) | 0.07 | 0.055 | 0.17 | |
| Use of glucarpidase | 7 | (2) | 3 | (0.01) | 4 | (4) | 0.08 | 0.75 | 0.68 | |
| Best response to first-line | 93 | |||||||||
| Complete remission | 224 | (66) | 171 | (68) | 53 | (60) | 0.19 | < 0.001 | < 0.001 | |
| Partial remission | 24 | (7) | 19 | (8) | 5 | (6) | 0.64 | 0.98 | 0.27 | |
| Refractory | 93 | (27) | 63 | (25) | 30 | (34) | 0.11 | < 0.001 | < 0.001 | |
| LOS, days (IQR) | 100 | 6 | (5–16) | 5.5 | (4–7) | 31 | (22–49) | < 0.001 | < 0.001 | 0.004 |
| Discharged to home, n (%) | 100 | 207 | (57) | 201 | (74) | 6 | (6) | < 0.001 | < 0.001 | 0.018 |
Abbreviations: KPS Karnofsky performance score, IQR interquartile range, LDH plasma lactico-deshydrogenase level, LOS length of hospital stay from first day of chemotherapy to discharge, MSKCC Memorial Sloan-Kettering Cancer Center score, MPV-A methotrexate, procarbazine, vincristine, and prednisone backbone (cytarabine or other consolidation therapies allowed), MTX methotrexate
At time of diagnosis, as compared with the LOS ≤ 15d group, patients of the LOS > 15d group presented more frequently with fever, sensory or motor deficit, gait disturbance, and/or cognitive impairment (Table 1). No differences in comorbidities were found (Supplemental Table 1). Patients LOS > 15d group had a lower KPS (P < 0.001), more frequent elevated cerebrospinal fluid (CSF) protein levels and a deep cerebral structure involvement, affecting mostly the brainstem and the thalamus.
Figure 1C depicts the patient path before the first day of chemotherapy. Briefly, patients of the LOS > 15d group had a longer time from admission to diagnostic confirmation (P < 0.001), contrasting with a shorter time from diagnosis to chemotherapy (P < 0.001). Patients from the LOS > 15d group were more likely to remain inpatient between diagnosis confirmation and chemotherapy. There was no difference in their admission patterns in intensive care unit, continuous care, emergency room, or community hospital (Supplemental Table 2). Patients from the LOS > 15d group were more frequently admitted first to a neurological ICU (P = 0.039) or a neurology unit (P = 0.01) whereas those in the LOS ≤ 15d group were more frequently admitted first to a neurosurgery unit (P < 0.001). There were no differences in the diagnostic confirmation procedure type between the two groups (Supplemental Table 2).
Concerning treatment, similar rates of radiotherapy, corticosteroids, hypertonic serum, or rituximab use were observed. We found a similar frequency of delayed methotrexate elimination or glucarpidase use when methotrexate concentrations were above expected values to prevent organ damage. Of note, no difference in response achievement was observed between the two groups (Table 1).
Patients of the LOS > 15d group were less prone to be discharged to their home (P < 0.001), as compared to another facility.
With a median follow-up of 46 months (IQR[27; 66]), median PFS was 9.3 months in the LOS > 15d group and 18 months in the LOS ≤ 15d group (P = 0.004, Supplemental Fig. 2). Median OS was 20 months in the LOS > 15d group and 62.2 months in the LOS ≤ 15d group (P < 0.001, Fig. 1D).
Variables significantly associated with PFS in univariate analysis were age (as a continuous variable per year increment), CSF protein level (as a continuous variable), epilepsy at diagnosis, and LOS. Variables significantly associated with OS in univariate analysis were age, KPS, MSKCC score, epilepsy at diagnosis, gait disturbance at diagnosis, behavior disorder at diagnosis, ocular involvement, and LOS. Deaths attributed to lymphoma were significantly more frequent in the LOS > 15d group (45% vs 30%, P = 0.012).
A multivariate Cox model analysis identified the following factors to be independently associated with PFS: LOS > 15d (HR:1.08, 95%CI[1.3–2.5], P < 0.001), epilepsy at diagnosis (HR: 2.13, 95%CI[1.28–3.54], P = 0.003), behavioral disorder at diagnosis (HR: 1.84, 95%CI[1.17–2.9], P = 0.009) and ocular involvement (HR: 2.06, 95%CI[1.31–3.24], P = 0.002). Factors associated with OS were age (HR: 1.04, 95%CI[1.01–1.07], P = 0.008), LOS > 15d (HR: 1.88, 95%CI[1.24–2.85], P = 0.003), and epilepsy at diagnosis (HR: 2.37, 95%CI[1.30–4.42], P = 0.005). Of note, other factors such as elevated CSF protein levels, previous admission in a ICU, or KPS at diagnosis were not significantly associated with OS or PFS in the multivariate analysis. Interestingly, results of the statistical analysis were similar with the 23-days threshold. As KPS and LOS may be probable cofounding factors, we conducted an ad hoc survival analysis depending on the LOS for each KPS stratum. Sixty percent of patients in the cohort had baseline KPS between 60 and 80%. LOS was still associated with OS and PFS in the 60% and 70% KPS strata, but not the 80% stratum (Supplemental Fig. 3). To further explore the interaction between KPS and LOS, we performed an interaction analysis between KPS and LOS, as well as a propensity score–adjusted analysis accounting for age, performance status, and neurological status. These analyses did not reveal a statistically significant effect of length of stay on survival, suggesting that the observed association partly—but not exclusively—reflects patients’ functional status.
Although epilepsy was associated with poorer OS and PFS, the underlying mechanisms remain unclear and may not be fully explained by antiepileptic drug interactions, [11] or by the cortical localization of the lesions.
Behavioral disorders were associated with poorer OS, probably reflecting deep cerebral localizations. 10 In our cohort the anatomic localization of the lymphoma lesions was not associated with survival [12].
An association of ocular involvement with poorer PFS has already been described, [13] the main hypothesis being the difficulty to achieve therapeutic concentrations of MTX in the vitreous.
Our study had several limitations, including its retrospective design and some date approximations. There was no review of imaging and a CSF workup was not performed on several patients. LOS may also be explained by several other patient-related factors such as hematologic tolerance, transfusion needs, or neutropenic infections, as well as local centre-related factors, family factors and other specific drivers of LOS such as rehabilitation needs or discharge planning issues. These factors are not well captured in the registry, or in the patient files. Therefore, the impact of LOS on prognosis can only be speculative and hypothesis-generating.
However, this cohort extracted from a national prospective registry used is large, homogeneous in age and treatment, allowing for a representative description of the patient hospital path with a long follow-up.
Altogether, this study indicates that a LOS > 15 days is independently associated with poorer survival outcomes in elderly PCNSL patients treated with MPV-based chemotherapy. While this finding highlights LOS as a potential prognostic marker, its underlying drivers require further investigation in larger and prospective cohorts.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors are particularly grateful to Jennifer Guilloteau for invaluable assistance in the data collection and verification.
Author contributions
RM: data analysis, manuscript writing. CH: data provision, registry data extraction, critical review of the manuscript. EG: data analysis, manuscript revision. ME: work through data provision, critical review and final approval of the manuscript. OC: work through data provision, critical review and final approval of the manuscript. PS: work through data provision, critical review and final approval of the manuscript. CS: work through data provision, critical review and final approval of the manuscript. RH: work through data provision, critical review and final approval of the manuscript. GA: work through data provision, critical review and final approval of the manuscript. VD: work through data provision, critical review and final approval of the manuscript. AC: work through data provision, critical review and final approval of the manuscript. LO: work through data provision, critical review and final approval of the manuscript. RU: work through data provision, critical review and final approval of the manuscript. HS: work through data provision, critical review and final approval of the manuscript. JA: work through data provision, critical review and final approval of the manuscript. OH: work through data provision, critical review and final approval of the manuscript.
Funding
Association Tours Autogreffe.
Data availability
No datasets were generated or analysed during the current study.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Gonzalez-Aguilar A et al (2012) Recurrent mutations of MYD88 and TBL1XR1 in primary central nervous system lymphomas. Clin Cancer Res 18(19):5203–5211. 10.1158/1078-0432.CCR-12-0845 [DOI] [PubMed] [Google Scholar]
- 2.Hernández-Verdin I et al (2023) Molecular and clinical diversity in primary central nervous system lymphoma. Ann Oncol 34(2):186–199. 10.1016/j.annonc.2022.11.002 [DOI] [PubMed] [Google Scholar]
- 3.Houillier C et al (2020) Management and outcome of primary CNS lymphoma in the modern era: an LOC network study. Neurology 94(10):e1027–e1039. 10.1212/WNL.0000000000008900 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ferreri AJM (2024) Primary central nervous system lymphomas: EHA–ESMO clinical practice guideline for diagnosis, treatment and follow-up. HemaSphere 8(6):e89. 10.1002/hem3.89 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ferreri AJM (2003) Prognostic scoring system for primary CNS lymphomas: the international extranodal lymphoma study group experience. J Clin Oncol 21(2):266–272. 10.1200/JCO.2003.09.139 [DOI] [PubMed] [Google Scholar]
- 6.Abrey LE et al (2006) Primary central nervous system lymphoma: the memorial Sloan-Kettering cancer center prognostic model. J Clin Oncol 24(36):5711–5715. 10.1200/JCO.2006.08.2941 [DOI] [PubMed] [Google Scholar]
- 7.Steward JS et al (2017) Effect of single agent high-dose methotrexate-related acute kidney injury on length of hospitalization and relative dose intensity in adult patients with central nervous system lymphoma. J Oncol Pharm Pract 23(7):496–501. 10.1177/1078155216665244 [DOI] [PubMed] [Google Scholar]
- 8.Demiralp B et al (2019) Length of stay, mortality, and readmissions among medicare cancer patients treated with glucarpidase and conventional care: a retrospective study. Clin Outcomes Res CEOR 11:129–144. 10.2147/CEOR.S188786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Candrilli S, Bell T, Irish W, Morris E, Goldman S, Cairo MS (2008) A comparison of inpatient length of stay and costs among patients with hematologic malignancies (excluding Hodgkin disease) associated with and without acute renal failure. Clin Lymphoma Myeloma 8(1):44–51. 10.3816/CLM.2008.n.003 [DOI] [PubMed] [Google Scholar]
- 10.Abrey LE et al (2005) Report of an international workshop to standardize baseline evaluation and response criteria for primary CNS lymphoma. J Clin Oncol 23(22):5034–5043. 10.1200/JCO.2005.13.524 [DOI] [PubMed] [Google Scholar]
- 11.Ferreri AJM et al (2004) Area under the curve of methotrexate and creatinine clearance are outcome-determining factors in primary CNS lymphomas. Br J Cancer 90(2):353–358. 10.1038/sj.bjc.6601472 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Jahr G, Broi MD, Holte H, Beiske K, Meling TR (2018) Evaluation of Memorial Sloan-Kettering Cancer Center and International Extranodal Lymphoma Study Group prognostic scoring systems to predict overall survival in intracranial primary CNS lymphoma. Brain Behav 8(3):e00928. 10.1002/brb3.928 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kreher S et al (2015) Prognostic impact of intraocular involvement in primary CNS lymphoma: experience from the G-PCNSL-SG1 trial. Ann Hematol 94(3):409–414. 10.1007/s00277-014-2212-z [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
No datasets were generated or analysed during the current study.