Edward R. Laws, Ian F. Parney, Wei Huang, Fred Anderson, Angel M. Morris, Anthony Asher, Kevin O. Lillehei, Mark Bernstein, Henry Brem, Andrew Sloan, Mitchel S. Berger, Susan Chang and Glioma Outcomes Investigators
Object. The Glioma Outcomes Project represents a contemporary analysis of the management of malignant (Grade III and Grade IV/GBM) gliomas in North America. This observational database was used to evaluate the influence of resection, as opposed to biopsy, on patient outcome as measured by the length of survival. Attempts were made to reduce the impact of selection bias by repeating the data analysis after omitting patients with major negative prognostic factors.
Methods. Outcome data from 788 patients accrued from multiple sites over a 4-year period (1997–2001) were analyzed with the primary outcome measure being length of survival. Of these, 565 patients with recent diagnoses formed the basis of the present analysis. Patients were systematically followed up until death or up to 24 months after enrollment in the study, and survival data were correlated with the histopathological grade and location of the tumor, the extent of surgery, the patient's performance status, and demographic factors.
The median length of survival was 40.9 weeks for patients with recently diagnosed GBMs. The true median length of survival for patients with Grade III gliomas was not reached, although there was a 58% survival rate at 104 weeks. In multivariate analysis, resection rather than biopsy (p < 0.0001), age 60 years or younger (p < 0.0001), and a Karnofsky Performance Scale (KPS) score of 70 or greater (p = 0.0004) were associated with a prolonged survival time for patients with Grade III or IV gliomas. The prognostic value of resection compared with biopsy was maintained (p < 0.0001), even after eliminating patients considered to be “poor risk” (those with age > 60 years, KPS score < 70, or presence of multifocal tumors), who may have been overrepresented in the biopsy group. Survival “tails” at 24 months were 58% for Grade III gliomas and 11% for GBMs.
Conclusions. These data provide Class II evidence to support tumor grade, patient's age, and patient's functional status as prognostic factors for survival in individuals with recently diagnosed malignant gliomas. Resection (compared with biopsy) is also a strong prognostic factor; however, no quantitative attempt was made to assess the true extent of the resection.
Edward F. Chang, Aaron Clark, Randy L. Jensen, Mark Bernstein, Abhijit Guha, Giorgio Carrabba, Debabrata Mukhopadhyay, Won Kim, Linda M. Liau, Susan M. Chang, Justin S. Smith, Mitchel S. Berger and Michael W. McDermott
Medical and surgical management of low-grade gliomas (LGGs) is complicated by a highly variable clinical course. The authors recently developed a preoperative scoring system to prognosticate outcomes of progression and survival in a cohort of patients treated at a single institution (University of California, San Francisco [UCSF]). The objective of this study was to validate the scoring system in a large patient group drawn from multiple external institutions.
Clinical data from 3 outside institutions (University of Utah, Toronto Western Hospital, and University of California, Los Angeles) were collected for 256 patients (external validation set). Patients were assigned a prognostic score based upon the sum of points assigned to the presence of each of the 4 following factors: 1) location of tumor in presumed eloquent cortex, 2) Karnofsky Performance Scale (KPS) Score ≤ 80, 3) age > 50 years, and 4) maximum diameter > 4 cm. A chi-square analysis was used to analyze categorical differences between the institutions; Cox proportional hazard modeling was used to confirm that the individual factors were associated with shorter overall survival (OS) and progression-free survival (PFS); and Kaplan–Meier curves estimated OS and PFS for the score groups. Differences between score groups were analyzed by the log-rank test.
The median OS duration was 120 months, and there was no significant difference in survival between the institutions. Cox proportional hazard modeling confirmed that the 4 components of the UCSF Low-Grade Glioma Scoring System were associated with lower OS in the external validation set; presumed eloquent location (hazard ratio [HR] 2.04, 95% CI 1.28–2.56), KPS score ≤ 80 (HR 5.88, 95% CI 2.44–13.7), age > 50 years (HR 1.82, 95% CI 1.02–3.23), and maximum tumor diameter > 4 cm (HR 2.63, 95% CI 1.58–4.35). The stratification of patients based on scores generated groups (0–4) with statistically different OS and PFS estimates (p < 0.0001, log-rank test). Lastly, the UCSF patient group (construction set) was combined with the external validation set (total of 537 patients) and analyzed for OS and PFS. For all patients, the 5-year survival probability was 0.79; the 5-year cumulative OS probabilities stratified by score group were: score of 0, 0.98; score of 1, 0.90; score of 2, 0.81; score of 3, 0.53; and score of 4, 0.46.
The UCSF scoring system accurately predicted OS and PFS in an external large, multiinstitutional population of patients with LGGs. The strengths of this system include ease of use and ability to be applied preoperatively, with the eventual goal of aiding in the design of individualized treatment plans for patients with LGG at diagnosis.