How often are nonenhancing supratentorial gliomas malignant? A population study

Methods.

We reviewed the records of all southern Alberta residents who were diagnosed with a glial neoplasm between January 1, 1994, and May 31, 1999, as recorded in the Alberta Cancer Registry, local hospital records, and the region’s brain tumor databases at the Tom Baker Cancer Center and Alberta Children’s Hospital. The inclusion criteria required a histologically proven supratentorial glioma in an adult or pediatric patient who had undergone preoperative contrast-enhanced CT or MRI. Classification of tumor anaplasia included low-grade glioma or subtypes (grades 1 and 2), anaplastic glioma or subtypes (grade 3), and glioblastoma multiforme (GBM, grade 4). Patients were then divided into four groups based on tumor grade (malignant vs low-grade) and contrast enhancement (present vs absent). As two groups were considered somewhat unusual, re-review of all available MRI studies for patients with nonenhancing malignant (15 of 21 cases) and contrast-enhancing low-grade gliomas (28 of 34) showed no discordance with the imaging reports.

Statistical analyses were carried out according to the Statistical Analysis System (SAS Institute Inc., SAS/STAT User’s Guide, Version 6, Fourth Edition, Cary, NC) and data expressed as mean ± SEM. Logistic regression related the risk of anaplasia in nonenhancing lesions to patient age. The operating characteristics (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV]) of the imaging studies were calculated. As this was a population-based study, the PPV and NPV could be calculated directly.

Results.

There were 314 patients (203 men and 111 women) identified with a supratentorial glioma, including 243 malignant and 71 low-grade gliomas. Age ranged from 4 months to 87 years. Mean age was 50 years. There were 267 patients evaluated with preoperative MRI and 47 patients who only had a preoperative CT scan.

Most malignant gliomas (222/243, 91%) demonstrated contrast enhancement, and only 21 (9%) malignant gliomas lacked enhancement. The mean patient age for the enhancing malignant gliomas was 56.5 ± 1.0 years (range, 10–86), compared to 43.6 ± 3.7 years (range, 17–81; p = 0.003, t-test) for nonenhancing malignant gliomas. The majority of contrast-enhancing malignant gliomas were GBM (165 patients, 74.3%), with smaller numbers of anaplastic astrocytomas (38, 17.1%), mixed gliomas (10, 4.5%), and oligodendrogliomas (9, 4.0%). In contrast, almost all of the nonenhancing malignant gliomas (figure 1) were anaplastic astrocytomas (18, 85.7%) with only a few anaplastic mixed gliomas (2, 9.5%) or oligodendrogliomas (1, 4.8%). In this study, the probability that a contrast-enhancing glioma was malignant was high (PPV = 86.7%), although almost 37% of the nonenhancing gliomas were also malignant (NPV = 63.8%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Left frontal lobe malignant astrocytoma in a 17-year-old woman. (A) Axial T2-weighted (3000/100) MRI shows an infiltrative hyperintense mass involving the parasagittal left frontal lobe. Coronal T1-weighted (600/8) (B) pre- and (C) postgadolinium images show no tumor enhancement.

In contrast to malignant gliomas, low-grade gliomas enhanced (34/71, 46.5%) about as commonly as they lacked enhancement (37/71, 53.5%; table). Patients with enhancing low-grade gliomas were younger (mean age, 27.4 ± 3.2 years) than those with nonenhancing low-grade tumors (mean age, 35.9 ± 1.7 years; p = 0.023, t-test). The group of enhancing low-grade gliomas was diverse and included many different subtypes (see the table), whereas the nonenhancing low-grade gliomas consisted almost exclusively of well-differentiated astrocytomas, oligodendrogliomas, and mixed gliomas.

We found a trend for an increased probability of malignancy with age (figure 2). In fact, the mean patient age for those with malignant nonenhancing gliomas was 7.7 years older than those with low-grade gliomas that lacked enhancement (p = 0.07, t-test).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. This graph shows the probability that a supratentorial nonenhancing lesion will be histologically malignant (i.e., anaplastic glioma or glioblastoma multiforme) as a function of age. The solid line represents the maximum likelihood probability according to a logistic regression model. Dashed lines represent 95% confidence bounds for the probability shown by the solid line. The dots are the individual patients with low-grade (probability of malignancy = 0.0) or malignant (probability of malignancy = 1.0) gliomas.

Discussion.

We found that contrast enhancement is not always helpful, as 9% of malignant gliomas lacked enhancement and almost 48% of low-grade gliomas enhanced. A “watch and wait” approach, without a histologic diagnosis, may therefore be inappropriate in a patient with a surgically accessible nonenhancing glioma as it could delay effective treatment in patients with malignant gliomas, particularly in older patients. We found a positive trend between the risk of anaplasia in nonenhancing gliomas and advancing age, similar to others.6 This concept is useful clinically but, in isolation, does not predict malignancy as considerable overlap in age exists between patients with malignant and nonmalignant gliomas. Other imaging modalities, such as MR spectroscopy, have promise in assessing tumors preoperatively but remain unproven in differentiating tumor subtype/grade to the level of confidence required to replace a tissue diagnosis.

Our study is population-based and the results should not be influenced by the selection biases in institutional series. We found that 9% of nonenhancing gliomas were malignant whereas others have reported higher rates (14–45%).2,4-7⇓⇓⇓⇓ It is possible that referral bias and the older age group6 may have contributed to the increased incidence in these series as any single patient would be more likely to have a malignant tumor.

We found that the group of low-grade glial neoplasms included histologic subtypes that typically enhance, such as gangliogliomas, pilocytic astrocytomas, pleomorphic xanthoastrocytomas, and subependymal giant cell astrocytomas. This group can sometimes be differentiated from malignant neoplasms based on location, young patient age, clinical presentation, or association with a neurocutaneous syndrome. We were surprised, however, to find 16 “ordinary” enhancing low-grade astrocytomas and oligodendrogliomas, although such tumors can approach 40 to 50% in some series.8-10⇓⇓ Variability in enhancement patterns likely also exists between pediatric and adult low-grade gliomas, which were combined in our study.

Our study has several limitations. Its retrospective nature is limited to existing documentation and available imaging studies that were incomplete in several cases. Although not all of our patients underwent MRI, all malignant nonenhancing gliomas did have preoperative MRI scans and our basic conclusion is therefore left intact. A few patients with contrast-enhancing low-grade and nonenhancing malignant glioma (five cases each) had only stereotactic biopsies, which are associated with sampling error and may have resulted in improper tumor grading. This issue may be more problematic in older patients who already have an increased risk of malignancy. Nonbiopsied patients with nonenhancing presumed low-grade gliomas were excluded from our study. This watchful waiting group is, by definition, more likely to be stable and would lower the incidence of nonenhancing malignant gliomas in our study if they truly represent low-grade gliomas. Finally, our description of enhancement as being either present or absent is likely overly simplistic but, unfortunately, is limited by our study’s retrospective nature. These last two points could be assessed in a prospective study in which all masses identified on imaging have objective grading of tumor enhancement prior to eventual biopsy.