Neoplastic Meningitis–Related Prognostic Significance of the Karnofsky Performance Status

Background  The prognostic significance of Karnofsky performance status in neoplastic meningitis (NM) has not been demonstrated in patient groups similarly matched for known prognostic variables.

Objective  To determine the effect of performance status on survival in NM.

Design  Retrospective comparison.

Setting  A university tertiary medical center.

Patients  Two well-matched cohorts with cytologically positive NM with (n = 30; group A) and without (n = 30; group B) independence in activities of daily living as defined by a Karnofsky performance status score of 70 or greater or less than 70, respectively.

Main Outcome Measures  Groups were matched on age, primary tumor, site of NM disease (cranial nerves or spinal cord), treatment (radiotherapy and chemotherapy; systemic and intraventricular), and absence of cerebrospinal fluid compartmentalization, NM-related encephalopathy, and neuroradiographic bulky central nervous system disease. Primary tumor histologic diagnoses included breast cancer (20 patients), non-Hodgkin lymphoma (10 patients), lung cancer (10 patients), melanoma (8 patients), and others (12 patients). At presentation, NM revealed cranial neuropathy (30 patients) or spinal cord dysfunction (39 patients). Radiotherapy was administered to 49 patients (whole brain only in 12 patients; restricted spine only in 35; whole brain and restricted spine in 2). All the patients received intraventricular chemotherapy, and 49 received concurrent tumor-specific systemic chemotherapy.

Results  Median survival was 6 weeks (range, 3-10 weeks) in group B compared with 15.5 weeks (range, 8-58 weeks) in group A (P < .001). No treatment-related deaths were observed. All the patients demonstrated progressive disease and died of either NM or systemic cancer.

Conclusions  A low Karnofsky performance status score predicts poor survival in patients with NM. Patients with a low Karnofsky performance status score may be best served by offering supportive care.

Neoplastic meningitis (NM) is a frequent complication of systemic cancer that occurs in approximately 5% of all patients with cancer.^1-4 However, deciding which patients to treat is problematic for several reasons. In nearly three-quarters of patients with NM, evidence of progressive systemic cancer is apparent.^1-4 Furthermore, one-third of patients with NM have coexistent bulky central nervous system (CNS) metastases defined by means of neuraxis neuroradiography, a previously documented predictor of poor survival odds in patients with NM.^1-4 In addition, another third of patients with NM have evidence of cerebrospinal fluid (CSF) compartmentalization by means of radioisotope ventriculography, another prognostic variable that predicts survival.³ In patients with persistent obstruction of CSF flow, survival is curtailed compared with patients without CSF obstruction. Last, in patients with NM-related encephalopathy (approximately 15% of all patients with NM), survival is curtailed compared with patients without encephalopathy. The present study compares patients with NM matched for recognized prognostic variables with respect to high (≥70 and independent in activities of daily living) and low (<70 and dependent in activities of daily living) Karnofsky performance status (KPS) scores. These results suggest that NM-related performance as measured using the KPS independently predicts survival in patients with NM.

This study, approved by the institutional review board of the University of South Florida's Moffitt Cancer Center in conjunction with the University of Massachusetts, comprised 2 groups of patients. Group B (n = 30) included all consecutive patients with a diagnosis of NM-related impaired KPS (defined as a KPS score of <70) seen between August 1, 1990, and January 31, 2007. During this period, 150 patients with a diagnosis of NM without KPS impairment (defined as a KPS score of ≥70) were seen. For each patient in group B (patients 31-60 in the Table), a patient from this cohort of 150 was matched with respect to age, sex, tumor histologic features, extent of disease (absent CSF compartmentalization and CNS bulky disease), site of NM disease (cranial nerve or spine), treatment, and absence of NM-related encephalopathy and composed group A (patients 1-30 in the Table).

The 60 patients (36 men and 24 women) ranged in age from 31 to 79 years (median age, 58 years). The KPS score at the time of diagnosis of NM in group B ranged from 40 to 60 (median, 60). The median KPS score in group A was 90, with a range of 70 to 100. All the patients had cytologically documented NM. In 4 patients (7%), NM was diagnosed at the time of initial systemic tumor presentation. In the 56 remaining patients, NM was diagnosed at 6 to 38 months (median, 16 months) after initial tumor presentation. Thirty patients had cranial nerve involvement (15 in each group) and 39 had NM-related spinal cord disease (19 in group B and 20 in group A).

In 20 patients (10 from each group), the primary tumor was in remission (ie, relapse manifested as isolated NM), and, therefore, only regional chemotherapy and limited-field CNS radiotherapy were used. In the remaining 40 patients (20 from each group), NM occurred in the context of either active or measurable systemic disease. In 40 patients (20 from each group), a variety of tumor-specific systemic chemotherapies were used in addition to regional chemotherapy and limited-field CNS radiotherapy. Seven patients in each group initially had evidence of CSF flow obstruction; however, after radiotherapy at the site of obstruction, normal CSF flow was restored in all.

All the patients underwent placement of an intraventricular catheter and reservoir, after which the extent of NM disease was evaluated as previously described.^1-4 All the patients received intraventricular chemotherapy (50 mg of liposomal cytosine arabinoside was used as the first intra-CSF agent in all patients) after the completion of involved-field radiotherapy as previously described.³

Contingency tables and the χ² test were used to summarize the relationship between patients' baseline characteristics and treatment with the presence of NM-related encephalopathy. One-way analysis of variance was used to examine for differences between group characteristics (age, sex, tumor histologic features, site of NM disease, extent of disease, and treatment). Kaplan-Meier plots and the log-rank test were used to evaluate the association of survival with baseline characteristics, treatments, and the presence of NM-related impairment in the KPS.^5-7

There were no significant differences among group characteristics of age, sex, tumor histologic features, site of NM disease, extent of disease, and NM-related treatment (P < .001). All 30 patients in group B were dependent in activities of daily living (KPS score of ≤60). Seven patients in group B underwent whole-brain radiotherapy, and 18 received involved-field spine radiotherapy. In contrast, all 30 patients in group A were independent in activities of daily living (KPS score of ≥70). Seven patients in group A were treated with whole-brain radiotherapy, and 19 received involved-field spinal cord radiotherapy.

No treatment-related deaths occurred. Sixteen of 60 patients (27%) developed grade 3 or 4 neutropenia or thrombocytopenia, of whom 6 required hospitalization for antibiotic drug treatment of neutropenic fever on at least 1 occasion, and 9 required platelet transfusion on at least 1 occasion. The CNS treatment-related toxic effects included intraventricular catheter infections (2 of 60 patients; 3%) and multiple episodes of chemically induced aseptic meningitis (32 of 60 patients; 53%).

Four patients (13%) in group B had brief responses to combined therapy, in contrast to 20 (67%) in group A. Survival differed between patients in groups A and B (P < .001). As seen in the Figure, patients in group B had a median survival of 6 weeks (range, 3-10 weeks; 95% confidence interval, 5.12-6.88 weeks). In contrast, patients in group A had a median survival of 15.5 weeks (range, 8-58 weeks; 95% confidence interval, 10.97-19.03 weeks). No patients survived for 3 or more months in group B, whereas 76.6% of patients survived for 3 months in group A (30.0% survived for 6 months and 13.0% for 1 year). Cause of death (determined by the investigator) differed between patient groups, with 80.0% of group B patients dying of progressive leptomeningeal disease compared with 40.0% of group A patients. These results were significant as assessed using χ² tests (P < .001).

Deciding which patients with NM to treat is difficult. Most patients have end-stage cancer, often have coexistent bulky CNS disease, and present with pleomorphic neurologic signs and symptoms, making disease recognition a challenge. Furthermore, because NM affects the entire neuraxis, evaluation requires craniospinal neuroradiography. Finally, optimum treatment of NM uses intraventricular chemotherapy, thereby requiring patients to undergo a neurosurgical procedure.^1-4,8-16 For all of these reasons, deciding which patients with NM to treat is clinically and economically relevant.

Several clinical characteristics define patients for whom standard NM-directed treatment is ineffective. These include patients with limited survival odds due to advanced systemic cancer and patients who, after discussion of the clinical implications of NM, decline further treatment. In addition, previous studies have suggested 3 other patient groups for whom NM-directed treatment is of limited benefit. In one group of patients, bulky subarachnoid or parenchymal CNS metastatic disease predicts a limited chance of survival. In the second group of patients, radiotherapy-resistant interruption of CSF flow, as defined by the study of radioisotope CSF flow, predicts a limited chance of survival.² In the third group, NM-related encephalopathy also predicts impoverished survival. The National Comprehensive Cancer Network guidelines regarding the treatment of NM stratify patients into good and poor risk groups.¹⁷ Poor risk is defined by a low KPS score, multiple serious fixed neurologic deficits, and extensive systemic disease with few treatment options. Using a low KPS score as a predictor of poor risk and, therefore, poor outcome in patients with NM is intuitively obvious but, to our knowledge, has never been formally evaluated. Jayson et al¹⁸ suggest that KPS (score of <70) predicts poor survival (median, 36 days) in a retrospective cohort of patients (n = 37) with breast cancer and carcinomatous meningitis. This study, however, did not formally evaluate other prognostic variables, ie, bulky neuroradiographic disease, CSF flow abnormalities, or carcinomatous encephalopathy known to affect survival in patients with NM. In addition, previous randomized trials have included only patients with KPS scores of 60 or greater, selecting, in effect, for good-risk patients with NM.

The results of the present study suggest another group of patients in whom NM-directed therapy may appropriately be deferred: those with NM-related impairment in performance as defined by a KPS score of 60 or less and in conjunction with the absence of carcinomatous encephalopathy, interruption of CSF flow, or bulky CNS metastasis. These results, admittedly based on a retrospective analysis and in a comparatively small number of patients, suggest that NM-related impairment in performance as defined by the KPS has prognostic significance and independently predicts survival in patients with NM who are considered for intra-CSF chemotherapy. By tailoring NM-directed therapies to appropriate patients with NM, physicians may more confidently provide palliative therapy for this challenging disease. The results of this study also highlight the need for more effective therapies for patients with NM and should encourage new therapeutic trials for this challenging category of patients.

Correspondence: Marc C. Chamberlain, MD, University of Washington, Department of Neurology, Fred Hutchinson Cancer Research Center, Seattle Cancer Care Alliance, 825 Eastlake Ave E, Mail Stop G-4940, Seattle, WA 98109-1023 (chambemc@u.washington.edu).

Accepted for Publication: August 26, 2008.

Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Chamberlain. Acquisition of data: Chamberlain. Analysis and interpretation of data: Chamberlain, Johnston, and Glantz. Drafting of the manuscript: Chamberlain and Johnston. Critical revision of the manuscript for important intellectual content: Chamberlain, Johnston, and Glantz. Statistical analysis: Chamberlain, Johnston, and Glantz. Administrative, technical, and material support: Chamberlain. Study supervision: Chamberlain.

Financial Disclosure: None reported.