|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 24, No 34 (December 1), 2006: pp. 5419-5426 © 2006 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.08.1497 Significance of Necrosis in Grading of Oligodendroglial Neoplasms: A Clinicopathologic and Genetic Study of Newly Diagnosed High-Grade Gliomas
From the Divisions of Neuropathology, Washington University School of Medicine, St Louis, MO; Mayo Clinic, Rochester, MN; and the University of Calgary, Calgary, Alberta, Canada Address reprint requests to Arie Perry, MD, Division of Neuropathology, Department of Pathology and Immunology, Washington University School of Medicine, 660 S Euclid, Campus Box 8118, St Louis, MO 63110; e-mail: aperry{at}wustl.edu
PURPOSE: High-grade gliomas (HGGs; WHO grades 3-4) are highly diverse, with survival times ranging from months to years. WHO 2000 grading criteria for high-grade oligodendroglial neoplasms [anaplastic oligoastrocytoma (AOA) and anaplastic oligodendroglioma (AO)] remain subjective, and the existence of grade 4 variants is controversial. PATIENTS AND METHODS: Overall survival (OS) of 1,093 adult patients with a cerebral HGG newly diagnosed between 1990 and 2005 was analyzed by univariate and multivariate models for significance of the following factors: patient age, surgery type, year of diagnosis, endothelial proliferation, necrosis, oligodendroglial histology, treatment center, and chromosome 1p, 19q, 7p (EGFR), and 10q (PTEN) abnormalities by fluorescence in situ hybridization (FISH).
RESULTS: Necrosis was a statistically significant predictor of poor OS on univariate and multivariate analyses in AOA but not in AO. Median OS for patients with necrotic AOA (22.8 months) was significantly worse than for patients with non-necrotic AOA (86.9 months; P < .0001) but was better than conventional glioblastomas (9.8 months; P < .0001). In addition to patient age, the following were significant independent prognostic factors (P CONCLUSION: Stratification of AOA, but not of pure AO, into grades 3 and 4 on the basis of necrosis is prognostically justified and is more powerful than the current approach. Both routine histology and genetic testing provide independent, prognostically useful information.
Diffuse, infiltrative gliomas are the most common primary intracranial neoplasms, accounting for 40% of all primary and 78% of all malignant CNS tumors.1 More than 80% of these tumors are considered high grade (grades 3 and 4) when diagnosed according to criteria established by WHO in 2000.2 Four distinct, diffuse, high-grade gliomas (HGGs) are currently recognized and are distinguished by morphologic evidence of differentiation along astrocytic [anaplastic astrocytoma (AA), WHO grade 3; and glioblastoma (GBM), WHO grade 4], oligodendroglial [anaplastic oligodendroglioma (AO), WHO grade 3], or mixed [anaplastic oligoastrocytoma (AOA), WHO grade 3] lineages. In contrast to the high-grade astrocytomas (HGAs), the existence of grade 4 high-grade oligodendroglial neoplasms (HGOs) remains a subject of debate. Classification and grading criteria for astrocytic neoplasms are well established, having been refined periodically over the last 80 years. WHO 2000 uses mitotic activity to distinguish AA from its WHO grade 2 counterpart. Endothelial proliferation (EP) and/or necrosis are distinguishing features of GBM. These morphologic criteria have enhanced reproducibility for HGA and serve as prognostically useful predictors of overall survival (OS).3-5 Since its original description,6 oligodendroglioma has become a well-recognized clinicopathologic entity, with increasing clinical interest in this important subclass over the last 20 years.7-9 The two forms of HGO, AO and AOA, are distinguished from their low-grade counterparts, oligodendroglioma and oligoastrocytoma, by the presence of any of the following histologic features: brisk mitotic activity, EP, and necrosis. Nonetheless, the grading remains subjective and is considerably less reproducible than for HGA,3-5,8,10 with survivals ranging widely from less than 1 to more than 10 years.2 Strong associations of chromosome 1p and 19q codeletion with therapeutic responsiveness and improved prognosis have made accurate histopathologic and molecular genetic characterization of HGO of paramount importance for proper treatment planning and enhanced prognostic accuracy.11-14 1p/19q codeletion most frequently (50% to 90%) occurs in morphologically classical AO. A subset (approximately 20%) of AOA also harbors 1p/19q codeletion,13 although its prognostic utility remains less well defined in this setting. Despite these advances, diagnostic discrepancies remain frequent.3-5,8,10 For example, small-cell astrocytoma (SCA) is considered a variant of GBM that commonly mimics HGO; however, its clinical behavior is similar to conventional GBM (median OS of approximately 1 year), even when EP and necrosis are not present.15,16 Identification of characteristic molecular genetic alterations in SCA, such as amplification of chromosome 7p (EGFR) and chromosome 10q (PTEN) deletion, is diagnostically useful in challenging patients. The present study was designed to address the natural history and prognostic factors of diffuse HGG diagnosed according to current WHO 2000 criteria. Genetic studies were performed in a subset of patients, and the main objective was to determine whether HGO could be further stratified histologically into grade 3 and 4 variants.
Study Participants With institutional review board approval (Washington University School of Medicine [WUSM] Human Studies Committee 04-1270), the surgical neuropathology and consult files were searched for primary HGGs diagnosed between 1990 and 2005. Patients less than 20 years of age and those having tumors located within the diencephalon, cerebellum, or spinal cord were excluded. Recurrent HGGs were also excluded to eliminate confounding effects of prior therapy, such as radiation necrosis. An additional 10 study patients with HGO with pseudopalisading necrosis (PPN) were obtained from the Mayo Clinic. In total, 1,093 adult patients with diffuse cerebral HGG were studied. Subsets of these patients have been previously reported.13,16-18 Information was extracted from medical records. Dates of death were obtained from these records or the Social Security Death Index. OS was calculated from the date of surgery to either the date of death or the last follow-up date documented in the patient's medical record.
Histopathology
Molecular Cytogenetics Dual-color fluorescence in situ hybridization (FISH) was performed and interpreted as previously described.13,17,18,20 In the vast majority of patients, the data were collected prospectively as part of the routine diagnostic work-up. Retrospective analysis was undertaken in selected patients who predated the routine application of FISH at WUSM in 1999 or were from the Mayo Clinic.
Statistics
Patients Characteristics of 1,093 study patients are listed in Table 1. Among the tumors, 13 (2.2% of GBMs or 1.2% of all HGGs) were gliosarcomas, and 120 (20.7% of GBMs or 11.0% of all HGGs) were SCAs. Two hundred fifteen tumors (48.8% of HGOs; 19.7% of all HGGs) were AOA. These latter two disproportionately high percentages reflected a consultation bias encountered by one of the authors (A.P.), given that SCA and AOA represented only 4.1% and 18.0%, respectively, of all in-house HGGs at WUSM versus 18% and 28.8%, respectively, of all consult HGGs (P < .001 for each). Growth of the consult service over time is also reflected in the significantly greater fraction of patients obtained after the year 2000 (71%; P  .001). More patients with HGO (AOA and AO) were diagnosed since 2000 (82% of 431 patients; P < .001), again likely reflecting a consultation bias because HGOs are more often submitted for second opinion and genetic analysis than HGAs. The mean age of GBM patients was 59.5 years and was significantly higher than the ages of patients with AA, AOA, or AO (P < .05). A significantly greater fraction of patients aged more than 60 years had GBM (52%; P < .05).
Roughly half of patients (555 patients; 51%) received their treatment at WUSM, and the GBM cohort was largely drawn from this group (428 patients; 74%). The majority of AOA and AO patients (72% and 85%, respectively) received care elsewhere. Most patients (76%) underwent resection; however, more HGA patients (32%) were diagnosed on stereotactic needle biopsy than HGO patients (7%; P < .05). Five hundred ninety-five patients (54%) died by the end of the study, and 58 (5.3%) were alive with longer than 5 years of follow-up. Mean follow-up time was 1.4 years for the entire cohort and varied with WHO 2000 diagnosis, patient age, diagnosis year, and treatment institution (Appendix Table A1, online only). Complete follow-up (C statistic = 100%) was obtained for 61% of patients, and 17% of patients were lost to follow-up after surgery (C 5%).
Histopathologic and Molecular Cytogenetic Features Molecular cytogenetic features are summarized in Appendix Figures A1 through A3 (online only). Decreasing frequencies of 1p/19q codeletion were found in AO (89%), AOA (19%), and AA/GBM (0%; Appendix Fig A2; P < .001). Polysomies (gains) of 1p or 19q were more frequent in HGA than in AO (data not shown; P < .001). EGFR amplification and 10q deletion were evident most frequently in GBM (Appendix Fig A3; P < .001). EGFR amplification was more frequent in SCA (63%) than other GBM subtypes (19%; P < .001). 10q deletion was identified in 82% of 141 HGAs but only 26% of 81 HGOs (P < .001). Additional information regarding the molecular cytogenetic features is available in the Appendix.
Univariate Analysis of Prognostic Factors
Both the presence of any type of necrosis (Fig 2A) and PPN alone (Fig 2B) were statistically significant negative prognostic factors only in patients with AOA (P .0001; Table 2). Median OS time for patients with necrotic AOA was significantly shorter (22.8 months; 95% CI, 14.9 to 33.8 months) compared with patients with non-necrotic AOA (86.9 months; 95% CI, 48.4 to 129 months). Similarly, median OS time for patients whose tumors featured PPN was significantly shorter (15.6 months; 95% CI, 10.5 to 30.1 months) compared with patients without PPN (62.8 months; 95% CI, 35.2 to 97.1 months). In contrast, neither necrosis nor PPN alone were significantly associated with OS in AO patients (P  .32; Figs 2C and 2D, Table 2). Small-cell histology did not significantly influence median OS time in GBM patients (11.5 v 9.4 months for SCA and conventional GBM, respectively; 95% CI, 9.7 to 14.1 and 8.3 to 11.0 months, respectively; P = .43).
As shown in Appendix Figure A5A (online only), trends were evident for 1p/19q codeletion in AOA (P = .12) and AO (P = .06) patients. Median OS time for patients with 1p/19q–deleted AOA was 132 months (95% CI, 29.5 months to upper limit undefined) compared with 42.0 months (95% CI, 33 to 86.9 months) for patients lacking 1p/19q codeletion. Similarly, median OS times for AO patients with and without 1p/19q codeletions were 135 months (95% CI, 76.0 months to upper limit undefined) and 68.2 months (95% CI, 26.3 months to upper limit undefined), respectively. However, only 19% of 189 AOA patients harbored 1p/19 codeletion, and 12% of 200 AO patients lacked 1p/19 codeletion, and few of these patients had died (13 of 58 patients; 22%), limiting the statistical power in these individual subsets. Therefore, a combined analysis of 1p/19q codeletion in all HGO patients was performed (Appendix Fig A5B) and showed significantly prolonged median OS time in 1p/19q codeleted HGO patients (132 months; 95% CI, 77.4 months to upper limit undefined) compared with patients without codeletion (49.6 months; 95% CI, 33.1 to 86.8 months; P < .0001).
Chromosomal abnormalities, including 1p, 19q, and 10q deletions, EGFR amplification, 1p/19q codeletion, and combined EGFR amplification/10q deletion, were each highly associated with OS for all HGG patients (P
Multivariate Analysis of Prognostic Factors
A multivariate comparison of the WHO 2000 versus modified grading schemes was conducted with the entire HGG cohort (Table 3). Age, grade, and surgery type were included in these two models, but cytogenetic factors were omitted because of limited availability of 1p/19q data (54% of all HGGs). All three factors were significant in both models (P < .001). Both models were equally accurate and accounted for similar percentages of variation in OS, but the hazard ratio (HR) for grade was greater for the modified approach (4.3; 95% CI, 3.4 to 5.6) than the WHO 2000 scheme (3.8; 95% CI, 3.0 to 4.8).
Kaplan-Meier curves using the WHO 2000 criteria are shown in Figure 3A. Median OS times of patients with GBM, AA, AOA, and AO were 9.8 months (95% CI, 8.9 to 11.3 months), 40.1 months (95% CI, 26.8 months to upper limit undefined), 41.9 months (95% CI, 33.1 to 86.9 months), and 86.8 months (95% CI, 74.5 months to upper limit undefined), respectively. Three statistically separable curves were evident because all pairwise comparisons among diagnostic entities were significant (P < .0001) except AA versus AOA (P = .73). In contrast, the modified scheme (Fig 3B) yielded four statistically separable survival curves (GBM, MOA4, AA, and AO; P
The significant survival advantage for patients with 1p/19q codeleted HGO (Appendix Fig A5B) prompted a multivariate analysis of age, grade (WHO 2000 v the modified scheme), oligodendroglial histology (HGO v HGA), and 1p/19q codeletion in the subset of genetically analyzed patients (n = 570). As shown in Table 3, age and 1p/19q codeletion were significant independent prognostic factors (P < .001). Modified grading was statistically significant (HR = 2.1; 95% CI, 1.4 to 3.3; P < .001), whereas WHO 2000 grading only showed a trend towards significance (HR = 2.1; 95% CI, 0.9 to 4.7; P = .073). Moreover, the presence of an oligodendroglial component (HGO v HGA) was significant when analyzed in the context of the modified scheme (HR = 0.6; 95% CI, 0.4 to 0.9; P = .026) but not the WHO 2000 scheme (HR = 0.8; 95% CI, 0.3 to 1.7; P = .49).
To our knowledge, this represents the largest retrospective analysis of both histopathologic and molecular genetic features of HGG since the publication of the WHO 2000 classification scheme. Patient demographics and genetic data are consistent with prior publications. As detailed in the Appendix, therapeutic records were not available for most of the non-WUSM patients; therefore, a potential weakness of this study is that the effects of individual therapies on OS could not be further evaluated. However, an important new finding was that necrosis stratified AOA into two groups with distinct OS, independent of age and 1p/19q status. Median OS time for AOA was significantly shorter for tumors with necrosis than for tumors without necrosis (P < .0001) but was still considerably longer than the OS time of patients with conventional GBM (P < .0001). This finding is consistent with results of a phase III clinical trial of adjuvant procarbazine, lomustine, and vincristine in newly diagnosed HGO patients, which also identified necrosis as an independent prognostic factor.21 In the present study, PPN was likewise found to be a significant predictor of poor OS. However, use of any type of necrosis was preferable compared with PPN for AOA grading for the following reasons: it was more prevalent (33% v 15%, respectively), yielded slightly greater diagnostic accuracy (C index, 0.72 v 0.70, respectively), accounted for more variation in OS (R2, 31% v 29%, respectively), and conformed to the approach currently used in grading HGA. Based on these data, we propose the following modified classification of mixed HGGs, which is similar to the current WHO 2000 classification of HGA: a grade 3 variant (AOA, MOA3) without necrosis and a grade 4 variant with necrosis (MOA4). MOA4 has a worse prognosis than conventional MOA3 (WHO 2000 AOA) but a better prognosis than standard GBM, even in the absence of 1p/19q codeletions. The optimal terminology for this tumor type is unclear, although GBM with oligodendroglial features is one possibility that is already in common use. In line with the notion that GBM is an astrocytic tumor, this term would imply a grade 4 variant with a mixture of both cell types. In contrast to AOA, neither necrosis nor PPN were found to be independent prognostic factors in AO. Similar findings were evident in a recently published retrospective series of 98 AOs.22 Although additional follow-up on the most recently diagnosed AO patients (2000 to 2005) will be required to make a definitive conclusion, these preliminary data suggest that such patients retain a favorable prognosis despite these otherwise ominous features and, in line with the current WHO scheme, support the existence of only a single high-grade pure oligodendroglioma (AO, WHO grade 3). In diagnostically challenging patients, genetic studies can be useful for distinguishing AO (1p/19q codeletion) from SCA (EGFR amplification, 10q deletion). Whereas the latter chromosomal abnormalities do not provide prognostic information independent of histology and grade, 1p/19q codeletion enhances both the diagnostic accuracy of AO and provides independent prognostic information. Of additional interest, patients in the current series with HGO lacking codeletion still had a more favorable outcome than patients with HGA of similar grade, emphasizing the importance of both accurate histologic classification and ancillary genetic testing. For example, the median OS time of 5.7 years (95% CI, 2.2 years to upper limit undefined) for AO without deletions was considerably higher than the 1 to 3 years reported previously.21,23 This difference supports the notion that OS estimates for HGO without deletions have been artificially lowered by the inadvertent inclusion of SCA.15,16 Additionally, our MOA3 and MOA4 cohorts without deletions survived longer (median OS time, 55 and 22.8 months, respectively; 95% CI, 33.4 to 97.1 and 14.9 to 42.0 months, respectively) than AA (median OS time, 40 months; 95% CI, 26.8 months to upper limit undefined) and conventional GBM patients undergoing resection (median OS time, 11.7 months; 95% CI, 10.2 to 13.2 months). In fact, these differences are even greater if one considers that the application of pure astrocytoma grading criteria to MOA3 would have resulted in an alternate diagnosis of GBM in 66% of patients due to the presence of EP (Table 1). These data argue against the opinions of some that oligodendroglial tumors lacking 1p/19q deletions are merely astrocytomas and that the diagnosis of a mixed glioma should be avoided whenever possible. Because independent prognostic value was noted for oligodendroglial histology on multivariate analysis of the modified grading cohort with FISH data (Table 3), this further supports the contention that cytogenetic features alone cannot replace histologic definitions of oligodendroglial differentiation, despite the inherent subjectivities of the latter.
Patients and Methods Study participants. Clinical data included age at diagnosis, sex, year of diagnosis, tumor location, treatment center (academic v nonacademic, and Washington University School of Medicine [WUSM] v other academic), and type of surgical procedure (stereotactic needle biopsy v open biopsy or resection). Therapeutic intent of the procedure was not examined. Patients who had available records almost uniformly underwent either postoperative radiation alone (fractionated external beam or stereotactic radiosurgery) or combined radiation and alkylating agent–based chemotherapy. Because detailed therapeutic records were not available in most of the consult patients, the effects of individual therapies on overall survival (OS) were not further evaluated in this study. Results of recently reported clinical trials of adjuvant procarbazine, lomustine, and vincristine indicating that the sequence of radiation and chemotherapy has no significant impact on median OS for patients with high-grade oligodendroglial neoplasms (HGOs) supports this approach (van den Bent MJ, Carpentier AF, Brandes AA, et al. J Clin Oncol 24:2715-2722, 2006; Cairncross G, Berkey B, Shaw E, et al. J Clin Oncol 24:2707-2714, 2006). Dates of death were obtained from the Social Security Death index (http://ssdi.rootsweb.com and http://www.familysearch.org). Histopathology. Histopathologic data included diagnosis, WHO 2000 grade, and presence or absence of endothelial proliferation (EP), geographic (ie, bland infarct-like) necrosis, and pseudopalisading necrosis (PPN). Ancillary immunohistochemical studies were performed in a subset of patients as necessary, with the most common markers being glial fibrillary acidic protein, S-100 protein, synaptophysin, neurofilament, Neu-N, p53, and Ki-67 (MIB-1). The authors considered a tumor to have oligodendroglial features when it displayed a major population of cells containing round uniform nuclei with either bland or vesicular chromatin, a sharply defined nuclear membrane, and often small nucleoli that increased in prominence in anaplastic foci. In lower grade areas, the cells generally had either minimal discernable cytoplasm, a small eccentric skirt of cytoplasm, or clear perinuclear haloes. In anaplastic foci, cells often acquired moderate quantities of amphophilic to eosinophilic cytoplasm imparting an epithelioid or plasmacytoid appearance. Other less specific features supporting oligodendroglial differentiation included mucin-rich microcysts, microcalcifications, prominent perineuronal satellitosis, subpial secondary structure formation, hypercellular nodules, and minigemistocytes. Nevertheless, the nuclear cytology represented the critical defining feature for all cell-type determinations of oligodendroglioma versus astrocytoma. Classic astrocytic cytology was defined by the presence of elongated, hyperchromatic nuclei with irregular nuclear contours. Tumor cells appeared as naked nuclei embedded within a dense fibrillary background (fibrillary), had discernable eosinophilic processes (fibrillary), eccentric bellies of eosinophilic cytoplasm (gemistocytic), large bizarre multinucleated forms (giant cell), or monomorphic oval nuclei with bland chromatin, minimal cytoplasm, and high mitotic index (small cell). Gliosarcomas had discernable islands of reticulin-rich, malignant-appearing mesenchymal elements. There is considerable diagnostic ambiguity regarding anaplastic oligoastrocytoma (AOA) within the neuropathology community (Aldape K, Simmons ML, Davis RL, et al. Cancer 88:2342-2349, 2000; Brat DJ, Castellano-Sanchez A, Kaur B, et al. Adv Anat Pathol 9:24-36, 2002; Coons SW, Johnson PC, Scheithauer BW, et al. Cancer 79:1381-1393, 1997; Perry A. Adv Anat Pathol 8:183-199, 2001; Prayson RA, Agamanolis DP, Cohen ML, et al. J Neurol Sci 175:33-39, 2000). For the purposes of this study, AOAs were defined by the presence of two major components displaying classic oligodendroglial and classic astrocytic cytologic features respectively as defined earlier, either in spatially separate (biphasic or compact) or mixed (intermingled or diffuse) patterns in accordance with the WHO 2000 classification scheme (Kleihues P, Cavenee WK: International Agency for Research on Cancer: Pathology and Genetics of Tumours of the Nervous System. Lyon, France, IARC Press, 2000). Small-cell astrocytomas (SCAs; n = 120) were diagnosed as previously described and, based on their natural history and molecular genetic features, were included with glioblastoma (GBM) for statistical purposes, unless otherwise indicated (Burger PC, Pearl DK, Aldape K, et al. J Neuropathol Exp Neurol 60:1099-1104, 2001; Perry A, Aldape KD, George DH, et al. Cancer 101:2318-2326, 2004). Diagnostic review. Original diagnoses were made by five staff neuropathologists at WUSM, although one of the authors (A.P.) was the primary (n = 156), secondary (n = 156), or consultant (n = 542) neuropathologist on 854 patients (78%). Diagnosis of the remaining 239 patients often reflected a consensus opinion in challenging patients, indicating the high level of intradivisional consultation routinely practiced in our institution. Study patients diagnosed before 2000 were rereviewed (94% of 315 patients) using current WHO criteria (Kleihues P, Cavenee WK: International Agency for Research on Cancer: Pathology and Genetics of Tumours of the Nervous System. Lyon, France, IARC Press, 2000). Patients diagnosed later were already classified using this scheme, although a subset of these patients (39% of 778 patients) was similarly rereviewed; these were mostly patients in whom the type of necrosis was not specified. Of all study patients, 1,074 (98%) were reviewed by A.P. (n = 558), C.R.M. (n = 220), or both authors (n = 296), and diagnoses of 105 patients (10%) were modified on rereview. Grading was clarified in 39 patients (37 high-grade astrocytoma [HGA] patients and two HGO patients); cytology was clarified in 66 patients, including 40 HGO patients originally diagnosed as HGA, 22 HGA patients originally diagnosed as HGO, and four AOA patients originally diagnosed as anaplastic oligodendroglioma (AO). Molecular cytogenetics. Paired probes included 1p32/1q42 (RPCI-11 Human BAC library clones 260I23 and 184E11, respectively; Invitrogen, Huntsville, AL), 19p13/19q13 (RPCI-11 clones 575H1 and 426G3, respectively; Invitrogen), CEP7 (Vysis Inc, Downers Grove, IL)/EGFR (7p12) (RPCI-11 clone 184P17; Washington University Human Genome Sequencing Center, St Louis, MO), and PTEN (10q23.3)/DMBT1 (10q25.3-q26.1) (provided by Dr Robert Jenkins, Mayo Clinic, Rochester, MN). Nuclei were counterstained with DAPI (Insitus, Albuquerque, NM). Green and red probe signals were enumerated under an Olympus BX60 fluorescence microscope with appropriate filters (Olympus, Melville, NY) on a minimum of 100 nonoverlapping nuclei.
Statistics.
Age was trichotomized as
Results
Molecular cytogenetic features. Molecular cytogenetic features of all high-grade gliomas (HGGs) are given in Figures A1 to  A3. 1p or 19q deletion was much more frequent in HGO than in HGA and, within the HGO, was more common in AO than in AOA (P < .001; Fig A1). 1p/19q codeletion and solitary 19q deletion were also more frequent in HGO than HGA, particularly for AO but also for AOA (P < .001; Fig A2). In contrast, EGFR amplification, 10q deletion, and combined EGFR amplification/10q deletion were much more frequent in GBM than in anaplastic astrocytoma (AA), AOA, or AO (P < .001; Fig A3). This was particularly true for the SCA variant. No significant differences in any of the seven examined molecular genetic features were evident between grade 3 mixed oligoastrocytoma (MOA3) and grade 4 mixed oligoastrocytoma (MOA4; P .22). However, all seven were significantly different between GBM and MOA4 (P .024). These relationships were particularly true for SCA versus MOA4 (P .004), but only 1p, 19q, 1p/19q, and 10q deletions were significantly different in non–small-cell GBM versus MOA4 (P .042).
Diagnostic utility of molecular cytogenetic features. Combined deletion of 1p and 19q (1p/19q codeletion) was evident in 89% of AO compared with 19% of AOA (P < .05; Fig A2). None of 181 AA/GBM patients showed 1p/19q codeletion. AOA patients more frequently exhibited deletion of 19q alone (solitary 19q deletion, 25% of 189 patients) than did AA, GBM, or AO patients (7% of 382 patients). Therefore, 1p/19q codeletion demonstrated relatively high sensitivity (89%) and specificity (91%) for the diagnosis of AO. The DOR for 1p/19q codeletion for AO was 74, indicating that a tumor featuring 1p/19q codeletion was 74 times more likely to be an AO than another HGG subtype. In contrast, solitary 19q was specific for AOA (93%) but not sensitive (25%). The DOR of solitary 19q deletion for AOA was only 5. EGFR amplification was specific (94%) but not sensitive (54%) for the diagnosis of GBM. Patients with EGFR amplification were 20 times more likely to be GBM than any of the other three diagnoses. The sensitivity and specificity of 10q deletion for a diagnosis of GBM were 81% and 71%, respectively; the DOR was 14. The sensitivity, specificity, and DOR of combined EGFR amplification and 10q deletion for GBM were 48%, 95%, and 16, respectively. Solitary 1p deletion was evident in only six patients, four of whom had HGA, including three SCAs. The diagnostic utility of molecular cytogenetic features in the context of GBM versus MOA4 is shown in Figure A4. Presence of 1p deletion or 1p/19q codeletion strongly favors a diagnosis of MOA4 in this context because MOA4s were more than 85 times more likely to harbor these abnormalities than were GBMs. In contrast, EGFR amplification, 10q deletion, and combined EGFR amplification/10q deletion favor the diagnosis of GBM; specifically, GBMs were 22, 21, and 31 times more likely to have these respective abnormalities than MOA4.
Discussion Our data confirm the independent prognostic significance of codeletion of chromosomes 1p and 19q in AO, although the statistical association (Cox P = .056; Table 3) was not as strong as those in previously published series (van den Bent MJ, Carpentier AF, Brandes AA, et al. J Clin Oncol 24:2715-2722, 2006; Cairncross G, Berkey B, Shaw E, et al. J Clin Oncol 24:2707-2714, 2006; Bauman GS, Ino Y, Ueki K, et al. Int J Radiat Oncol Biol Phys 48:825-830, 2000; Cairncross JG, Ueki K, Zlatescu MC, et al. J Natl Cancer Inst 90:1473-1479, 1998; Smith JS, Perry A, Borell TJ, et al. J Clin Oncol 18:636-645, 2000).
The median OS time of patients with AO harboring 1p/19q codeletion was longer (135 months; 95% CI, 76.0 to > 168 months) than that of AO without codeletion (68.2 months; 95% CI, 26.3 to > 168 months; log-rank P = .06). A similar effect was evident in AOA; the median OS time of patients with AOA harboring codeletion was longer (132 months; 95% CI, 29.5 to > 168 months) than for AOA patients without codeletion (42 months; 95% CI, 33.0 to 86.9 months). A trend towards statistical significance of 1p/19q codeletion in AOA was evident on univariate analysis (P = .12) but not on multivariate analysis (P The fact that only 19% of 389 HGO patients with known 1p/19q status had died by study end strengthens these conclusions and prompted us to reanalyze 1p/19q status in the entire HGG cohort for which 1p/19q had been determined (n = 570). 1p/19q codeletion was associated with significantly better OS in both univariate (P = .0001; Fig A5) and multivariate analyses (P < .001; Table 3).Furthermore, when HGOs were graded by modified criteria into three distinct entities (AO, MOA3, and MOA4), the presence of an oligodendroglial component (HGO v HGA) became a significant favorable prognostic factor (P = .026; Table 3) independent of age, modified grade, and 1p/19q codeletion status. These results suggest that both histology (HGO v HGA) and cytogenetics are independent prognostic factors in HGG.
It should be noted that 39 (48%) of 81 AA, 111 (52%) of 215 AOA, and 121 (56%) of 216 AO patients were from the most recent accrual group (2003 to 2005), based primarily on the growth of the consult service of one of the authors (A.P.) and the increased interest in 1p/19q molecular genetic testing of HGO over this time period. As a result, limited follow-up time was available for these patients, and few had died by study end. This limitation is reflected in the CIs for AA, AO (with or without 1p/19q codeletion), and AOA with 1p/19q codeletion, all of which have ambiguous upper limits. Accurate survival estimates for these cohorts will require continued follow-up. Molecular genetic analysis is a useful adjunct to histopathologic evaluation of HGG. Because diagnosis of an oligodendroglial component in mixed HGG can be subjective, molecular analyses may be useful in distinguishing MOA4 and GBM in patients featuring necrosis. In this regard, deletion of 1p, 19q, or both strongly supported the diagnosis of MOA4 versus GBM (Fig A4). In contrast, EGFR amplification, 10q deletion, or combined EGFR amplification/10q deletion favored the diagnosis of GBM versus MOA4 (Fig A4), although the magnitude of these differences (DOR, 21 to 32) was less dramatic than for the MOA4-related abnormalities (DOR, > 85). Thus, concurrent histopathologic evaluation and molecular genetic analysis may be useful in distinguishing MOA4 from GBM in difficult patients.
The authors indicated no potential conflicts of interest.
We thank Ruma Banerjee, Diane Robirds, and Derek Van Booven for technical assistance. We are grateful to the following investigators for provision of patient follow-up: Rob Macaulay (Dalhouise University, Halifax, Nova Scotia, Canada); Harker Rhodes (Dartmouth University, Lebanon, NH); Jane Barron (Health Care Corporation of St Johns, New Brunswick, Newfoundland, Canada); Jeff Winslow (Lancaster, PA); Daniel Brown (Lehigh Valley, Allentown, PA); Ravi Raghavan (Loma Linda Univervisty, Loma Linda, CA); Henry Brown (Loyola, Chicago, IL); Richard Hessler (Medical College of Georgia, Augusta, GA); Eileen Bigio, Nina Paleologos, and Numa Gottardi (Northwestern University, Chicago, IL); Chris Robinson (Royal University Hospital, Saskatoon, Saskatchewan, Canada); Elizabeth Cochran (Rush, Chicago, IL); Beth Levy (St Louis University, St Louis, MO); E.S. Jones (Southeast Regional Health Authority, Moncton, New Brunswick, Canada); Nancy Karpinski (St Alphonsus, Boise, ID); Terri Haddix (Stanford, Palo Alto, CA); Howard Chang (State University of New York Upstate, Syracuse, NY); Mark Curtis (Thomas Jefferson, Philadelphia, PA); Edward Johnson (University of Alberta, Edmonton, Alberta, Canada); Arthur Clark, Bernadette Curry, David George, and Lisa Steele (University of Calgary, Calgary, Alberta, Canada); Robert Wollman (University of Chicago, Chicago, IL); Anthony Yachnis (University of Florida, Gainesville, FL); Tibor Valyi-Nagi (University of Illinois at Chicago, Chicago, IL); Karen Santa Cruz (University of Minneapolis, Minneapolis, MN); John Woulfe (University of Ottawa, Ottawa, Ontario, Canada); Susan Eliason (University of South Dakota, Rapid City, SD); Claudia Greco (University of California Davis, Sacramento, CA); Steve Rostad (Washington Pathology Association, Seattle, WA); and Marc del Bigio (Winnipeg Health Sciences Center, Winnipeg, Manitoba, Canada).
Supported by Cancer Biology Training Grant No. T32CA009547 from the National Cancer Institute, Bethesda, MD (C.R.M.). Presented in part at the 42nd Annual Meeting of the American Society of Clinical Oncology, June 2-6, 2006, Atlanta, GA. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Central Brain Tumor Registry of the United States: Statistical Report: Primary Brain Tumors in the United States, 1998-2002. Hinsdale, IL, Central Brain Tumor Registry of the United States, 2006 2. Kleihues P, Cavenee WK: International Agency for Research on Cancer: Pathology and Genetics of Tumours of the Nervous System. Lyon, France, International Agency for Research on Cancer Press, 2000 3. Aldape K, Simmons ML, Davis RL, et al: Discrepancies in diagnoses of neuroepithelial neoplasms: The San Francisco Bay Area Adult Glioma Study. Cancer 88:2342-2349, 2000[CrossRef][Medline] 4. Coons SW, Johnson PC, Scheithauer BW, et al: Improving diagnostic accuracy and interobserver concordance in the classification and grading of primary gliomas. Cancer 79:1381-1393, 1997[CrossRef][Medline] 5. Prayson RA, Agamanolis DP, Cohen ML, et al: Interobserver reproducibility among neuropathologists and surgical pathologists in fibrillary astrocytoma grading. J Neurol Sci 175:33-39, 2000[CrossRef][Medline] 6. Bailey P, Cushing H: A Classification of the Tumors of the Glioma Group on a Histogenetic Basis With a Correlated Study of Prognosis. Philadelphia, PA, J.B. Lippincott Company, 1926 7. Burger PC: What is an oligodendroglioma? Brain Pathol 12:257-259, 2002[Medline] 8. Perry A: Oligodendroglial neoplasms: Current concepts, misconceptions, and folklore. Adv Anat Pathol 8:183-199, 2001[CrossRef][Medline] 9. Reifenberger G, Louis DN: Oligodendroglioma: Toward molecular definitions in diagnostic neuro-oncology. J Neuropathol Exp Neurol 62:111-126, 2003[Medline] 10. Brat DJ, Castellano-Sanchez A, Kaur B, et al: Genetic and biologic progression in astrocytomas and their relation to angiogenic dysregulation. Adv Anat Pathol 9:24-36, 2002[CrossRef][Medline] 11. Bauman GS, Ino Y, Ueki K, et al: Allelic loss of chromosome 1p and radiotherapy plus chemotherapy in patients with oligodendrogliomas. Int J Radiat Oncol Biol Phys 48:825-830, 2000[CrossRef][Medline] 12. Cairncross JG, Ueki K, Zlatescu MC, et al: Specific genetic predictors of chemotherapeutic response and survival in patients with anaplastic oligodendrogliomas. J Natl Cancer Inst 90:1473-1479, 1998 13. Perry A, Fuller CE, Banerjee R, et al: Ancillary FISH analysis for 1p and 19q status: Preliminary observations in 287 gliomas and oligodendroglioma mimics. Front Biosci 8:a1-a9, 2003[Medline] 14. Smith JS, Perry A, Borell TJ, et al: Alterations of chromosome arms 1p and 19q as predictors of survival in oligodendrogliomas, astrocytomas, and mixed oligoastrocytomas. J Clin Oncol 18:636-645, 2000 15. Burger PC, Pearl DK, Aldape K, et al: Small cell architecture: A histological equivalent of EGFR amplification in glioblastoma multiforme? J Neuropathol Exp Neurol 60:1099-1104, 2001[Medline] 16. Perry A, Aldape KD, George DH, et al: Small cell astrocytoma: An aggressive variant that is clinicopathologically and genetically distinct from anaplastic oligodendroglioma. Cancer 101:2318-2326, 2004[CrossRef][Medline] 17. Fuller CE, Schmidt RE, Roth KA, et al: Clinical utility of fluorescence in situ hybridization (FISH) in morphologically ambiguous gliomas with hybrid oligodendroglial/astrocytic features. J Neuropathol Exp Neurol 62:1118-1128, 2003[Medline] 18. Fallon KB, Palmer CA, Roth KA, et al: Prognostic value of 1p, 19q, 9p, 10q, and EGFR-FISH analyses in recurrent oligodendrogliomas. J Neuropathol Exp Neurol 63:314-322, 2004[Medline] 19. Giannini C, Scheithauer BW, Weaver AL, et al: Oligodendrogliomas: Reproducibility and prognostic value of histologic diagnosis and grading. J Neuropathol Exp Neurol 60:248-262, 2001[Medline] 20. Brat DJ, Seiferheld WF, Perry A, et al: Analysis of 1p, 19q, 9p, and 10q as prognostic markers for high-grade astrocytomas using fluorescence in situ hybridization on tissue microarrays from Radiation Therapy Oncology Group trials. Neurooncol 6:96-103, 2004[Abstract] 21. van den Bent MJ, Carpentier AF, Brandes AA, et al: Adjuvant procarbazine, lomustine, and vincristine improves progression-free survival but not overall survival in newly diagnosed anaplastic oligodendrogliomas and oligoastrocytomas: A randomized European Organisation for Research and Treatment of Cancer phase III trial. J Clin Oncol 24:2715-2722, 2006 22. Smith SF, Simpson JM, Brewer JA, et al: The presence of necrosis and/or microvascular proliferation does not influence survival of patients with anaplastic oligodendroglial tumours: Review of 98 patients. J Neurooncol 80:75-82, 2006[CrossRef][Medline] 23. Cairncross G, Berkey B, Shaw E, et al: Phase III trial of chemotherapy plus radiotherapy compared with radiotherapy alone for pure and mixed anaplastic oligodendroglioma: Intergroup Radiation Therapy Oncology Group Trial 9402. J Clin Oncol 24:2707-2714, 2006 Submitted July 4, 2006; accepted September 27, 2006.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 or Fisher's exact tests were used for comparisons of proportions, and Student's t test was used for mean comparisons of continuous variables such as age. All statistical tests were two sided, and P < .05 was considered significant unless otherwise stated. Kaplan-Meier survival curves were plotted, and log-rank tests and Cox proportional hazards analysis were used for univariate and multivariate comparisons of median OS. Additional methodologic details are available in the Appendix. 
