Gliomas of the Optic Nerve: A SEER-Based Epidemiologic Study

To determine whether patients with biopsy-confirmed optic nerve glioma differ in clinical features and outcomes from those diagnosed by neuroradiologic imaging alone.

Original Contribution Section Editors: Clare Fraser, MD Susan Mollan, MD Gliomas of the Optic Nerve: A SEER-Based Epidemiologic Study Fatma Dihowm, MD, MS, Luis A. Alvarado, MS, Curtis E. Margo, MD, MPH Background: To determine whether patients with biopsyconfirmed optic nerve glioma differ in clinical features and outcomes from those diagnosed by neuroradiologic imaging alone. Methods: Retrospective comparative analysis. Pilocytic astrocytomas (PAs) and gliomas of the optic nerve were identified through ICD-O codes in the Surveillance, Epidemiology, and End Results (SEER) cancer registry from 1975 through 2017. Demographics, clinical features, and outcomes were compared according to the method of diagnosis (biopsy-confirmed and radiologic only) and by age (birth through 19 years and 20 years of age and older). Differences in proportions were tested with the chi-square test. Associations with tumor-related death were evaluated with logistic regression. Statistical significance: a , 0.01. Results: Over 42 years, 313 PAs and 720 gliomas of the optic nerve were identified. The young age distributions were similar between the 2 groups. PAs were biopsied more often than gliomas (54% vs 13.2% [P , 0.001]). Tumorattributable death occurred more often among PAs and gliomas that were biopsied than those that were not (7.1% vs 0.7% [P , 0.01]; 7.4% vs 1.1% [P , 0.01], respectively). Roughly 15% of both PAs and gliomas were diagnosed in persons 20 years and older. Conclusions: Biopsy-confirmed cases of PA and glioma of the optic nerve were associated with more therapeutic interventions and worse outcomes compared with patients who were diagnosed radiologically. Clinical variables relevant to clinical decision-making not captured by SEER likely explain the inability to meaningfully interpret outcome from the registry Department of Internal Medicine (FD), Paul L. Foster School of Medicine, Tech University Health Sciences Center Texas, El Paso, Texas; Biostatistics and Epidemiological Consulting Lab (LAA), Paul L. Foster School of Medicine, Texas Tech University Health Sciences Center El Paso, Texas; and Department of Pathology and Cell Biology, and Ophthalmology (CEM), Morsani College of Medicine, University of South Florida, Tampa, Florida. The authors report no conflicts of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www. jneuro-ophthalmology.com). Address correspondence to Curtis E. Margo, MD, 13330 USF Laurel Dr., 4th Floor, Dermatopathology, Tampa, FL 33612; E-mail: [e-mail] 462 database. Cancer registries should avoid coding specific histopathologic diagnoses when tissue is not obtained. Journal of Neuro-Ophthalmology 2022;42:462–469 doi: 10.1097/WNO.0000000000001630 © 2022 by North American Neuro-Ophthalmology Society G liomas of the optic nerve are rare, and information regarding their epidemiology, diagnosis, and management is usually obtained from studies conducted at referral institutions or from surveys of the literature (1–6). The term glioma refers to a neoplasm derived from any of the various cells of interstitial tissue of the brain and spinal cord. Gliomas of the optic nerve are generally considered low-grade neoplasms, when dealing with pilocytic astrocytoma (PA). Authorities once debated whether they were hamartomas or benign neoplasms (1). Consensus opinion now views them as benign neoplasm, although capable of causing significant morbidity (7,8). Despite neurooncologic judgment of benignancy, the Surveillance, Epidemiology, and End Results (SEER) cancer registry had maintained a policy of reporting pilocytic/juvenile astrocytomas as behavior code/3 “malignant” until December 2018 when coding of optic nerve astrocytomas was changed to reflect World Health Organization (WHO) guidelines as nonmalignant (behavior code/1) (9). To the best of our knowledge, there has been only 2 studies of optic nerve gliomas using the SEER data set (10,11). One study excluded patients older than 19 years for the purpose of examining racial incidence (10). The other study examined the influence of tumor grade on outcome and excluded 71% of cases because of the absence of grading (11). We explored the SEER data set from 1975 through 2017 for optic nerve gliomas without limitation of age to compare outcomes based on whether they were diagnosed on tissue biopsy or through radiologic criteria alone. METHODS Patients with optic nerve glioma were obtained from the caselisting section of SEER databases of the National Cancer Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution Institute using the SEER 9 registry (1975–2017). Topography code C73.3 for the optic nerve was linked to ICD0 morphology codes for gliomas and astrocytomas: 9380 through 9460 and behavior codes 1, 2, and 3 (12). Basic demographic information on age, gender, and race were abstracted. Age was stratified birth to 1 year, 1–4 years, 5–9 years, then on 10-year intervals to 50 years of age and older. The method of diagnosis was documented as through biopsy or based on only radiologic criteria. Laterality was recorded as right, left, bilateral, or midline. Cases recorded as not a paired site were included as midline. Treatment was recorded as surgery (other than biopsy), radiation, and/or chemotherapy. The outcome was documented as alive or dead of other causes, death attributed to neoplasm, or not applicable if another tumor was known. Chi-square tests were used to test statistical significance of proportional differences at P , 0.01. The risk of tumor-attributable death was examined using penalized logistic regression (Firth logit) analysis. Odds ratios were reported with 99% confidence intervals (CIs) and adjusted for variables available to us in SEER: younger than 20 years, tumor biopsy performed, Caucasian race, male gender, unilateral location, surgical treatment, radiation therapy, and chemotherapy. A second adjusted model used the same dependent variables minus therapeutic interventions (surgery, chemotherapy, and radiation). Statistical analyses were conducted using Stata, version 15. Data were abstracted between November 1, 2020, and December 1, 2020 (12). The SEER cancer registry required all PAs and low-grade gliomas to be coded as malignant before 2018 (9). In this article, the term “malignant” will be avoided when speaking of gliomas of optic nerve that have traditionally been viewed as benign by histopathology (ICD-O codes 9380 and 9421). The ICD-O topography code for the optic nerve (C73.3) combines optic nerve, chiasm, and optic tract. The topography code neither distinguishes tumors of optic nerve proper from optic tract or chiasm nor can it distinguish subdivisions of the optic nerve (e.g., intrascleral, orbit, intracanalicular, or intracranial). The university IRB exempted this study as the SEER data are in the public domain and contain no protected personal health information. RESULTS ICD-O morphology codes for gliomas of the optic nerve revealed 1,158 cases from 1975 through 2017, of which 2 categories glioma (9380/3) and PA (9421/3) predominated, accounting for 89.9% (1,041 cases) of all cases (Table 1). Further analysis was restricted to these 2 groups because the remaining 8 categories had too few entries to meaningfully compare biopsy vs radiologic diagnoses. Pilocytic Astrocytoma Vs Glioma Information on age distribution, laterality, and treatment on 313 PAs and 720 gliomas diagnosed radiologically or Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 TABLE 1. Gliomas of the optic nerve* ICD-O Morphology Code† Total (%) Glioma, benign 9380/0 43 Glioma, borderline 9380/1 3 Glioma, malignant 9380/3 720 Glioma, mixed 9382/3 3 Astrocytoma, NOS 9400/3 42 Astrocytoma, 11 anaplastic 9401/3 Astrocytoma, fibrillary 9420/3 1 Astrocytoma, pilocytic 9421/3 321 Glioblastoma, NOS 9440/3 13 Gliosarcoma 9442/3 1 Total 1,158 Female (%) (3.7) 23 (53.5) (0.3) 1 (33) (62.2) 383 (53.2) (0.3) 1 (33) (3.6) 15 (35.7) (1.0) 6 (54.5) (0.1) 1 (100) (27.7) 156 (48.6) (1.1) 4 (30.8) (0.1) 1 (100) (100) 591 (51.0) *SEER topography code optic nerve C73.2 from 1975 to 2017. †Morphology code/3 for malignant required for all SEER central nervous system tumors regardless of histopathologic features until 2018. SEER, Surveillance, Epidemiology, and End Results. histologically are listed in Table 2. Of the 313 PAs, 152 (48.6%) occurred in girls or women, 256 (81.8%) were in Caucasians, and 269 (85.9%) occurred before the age of 20 (Table 2). Eight PAs were excluded because no information on either tissue or radiologic diagnosis was available. A slightly greater proportion of 720 gliomas of the optic nerve occurred in girls and women (383 [53.2%]), with similar majorities in Caucasians (601 [83.5%]) and persons younger than 20 years (611 [84.6%] (Table 2). Histologic confirmation of PA was made in 169 of 313 cases (54.0%) compared with just 95 of 720 gliomas (13.2%) (P , 0.01). There were substantial differences in surgical treatment (other than biopsy) recorded for 126 PAs (40.3%) compared with 28 gliomas (3.9%) (P , 0.01). Death attributed to neoplasm was noted in 13 PAs (4.2%) and in 14 gliomas (1.9%) (P , 0.052) (Table 2). The median follow-up was shorter among patients with PAs {69.0 months (interquartile [IQ] range: 28, 135)} compared with glioma (93.0 months [IQ range: 44.5, 152]) (P , 0.01). Pilocytic Astrocytoma Vs Glioma: Stratified by the Method of Diagnosis When cases were examined according to how diagnoses were established, there was a statistically significant larger proportion of deaths attributable to neoplasm in patients with PA who underwent biopsy (Table 3). Twelve attributable deaths (7.1%) occurred among those with a biopsy compared with 1 death (0.7%) when diagnosed radiologically (P , 0.01). The mean follow-up was longer in the biopsy group (113.5 months; IQ range 49.5, 179) compared with the radiologic cohort (42.5 months; IQ range 19, 77) (P , 0.01). Ninety-five tumors (13.2%) classified as optic nerve gliomas were diagnosed by biopsy (Table 4). There were 7 tumor-attributable deaths (7.4%) among those who underwent biopsy compared with 7 (1.1%) diagnosed 463 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 2. Major diagnostic classifications of optic nerve glioma SEER 1975–2017 Pilocytic Astrocytoma n = 313* (100%) Sex Female Race Caucasian Black Others or unknown Age Younger than 1 year 1–4 years 5–9 years 10–19 years 20–29 years 30–39 years 40–49 years 50 years and older Total Histologic diagnosis†‡ Laterality Right, single primary Left, single primary Bilateral, primary Not a paired site or midline§ Unknown or unspecified Treatment‡ Surgery Radiation (with or without surgery) Chemotherapy Outcomesk Alive or dead of other causes Death attributable to neoplasm Not applicable when other primary tumor known Follow-up (months)‡ Median (interquartile range) Glioma n = 720 (100%) 152 (48.6) 383 (53.2) 256 (81.8) 25 (8.0) 32 (10.3) 601 (83.5%) 59 (8.2%) 60 (8.4) 23 117 81 48 12 9 11 12 313 169 (7.3) (37.4) (5.9) (15.3) (3.8) (2.9) (3.5) (3.8) (100) (54.0) 45 302 165 99 27 24 33 25 720 95 (6.3) (41.9) (22.9) (13.8) (3.8) (3.3) (4.6) (3.5) (100) (13.2) 99 83 51 61 19 (31.6) (26.5) (16.3) (19.4) (6.1) 197 176 126 166 55 (27.4) (24.4) (17.5) (23.0) (7.7) 126 (40.3) 27 (8.6) 86 (27.5) 28 (3.9) 54 (7.4) 193 (26.8) 280 (89.5) 13 (4.2) 20 (6.4) 673 (93.5) 14 (1.9) 32 (4.4) 69.0 (28.0–135.0) 93.0 (44.5–152.0) *Eight cases excluded due to diagnostic methods unknown, lacking documentation of both microscopic and radiologic confirmations. Before 2018, SEER required pilocytic astrocytoma of the optic nerve to be coded as malignant (/3 behavior code). † Remaining 144 cases (46.0%) of pilocytic astrocytoma and 625 cases (86.8) of glioma were diagnosed by radiologic criteria. ‡ Statistically significant P , 0.01, Mann–Whitney U test. § Implies chiasm when listed at not a paired site. k Statistically significant ,0.01, chi-square test. SEER, Surveillance, Epidemiology, and End Results. radiologically (P , 0.01). The median follow-up was similar: 98 months (IQ range 46, 177) for tissue diagnosis and 92 months (IQ range 44, 149) for radiologic diagnosis. Considerably more therapeutic interventions (surgery, radiation, and chemotherapy) were performed in patients who underwent biopsy for PA and for glioma compared with those with radiologic diagnoses alone (Tables 3 and 4). Pilocytic Astrocytoma Vs Glioma: The Method of Diagnosis Stratified by the Age Group When biopsy and radiologic groups of PA and optic nerve glioma were examined by age (birth through 19 years and 20 years and older), both diagnostic groups received different therapies and had different tumor-attributable deaths according to the age group (Tables 5 and 6). Older 464 patients with either PA or glioma were more likely to have tumor-directed surgery and radiation, whereas younger patients were more likely to receive chemotherapy. Of the 44 patients 20 years or older with PA, there were 3 attributable deaths (6.8%), roughly half the rate in younger patients {10 attributable deaths ([3.7%] [P = 0.44])}, not a statistically significant difference. Percent attributable death among younger patients with glioma 0.7% (4 deaths) was significantly less than those of age 20 and older (9.2% [10 deaths] [P , 0.01]). Logistics Regression Analysis The univariate odds ratio for tumor-related mortality for optic nerve glioma revealed that age older than 20 years, cancerdirected surgical treatment, radiation, and positive histologic Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 3. Pilocytic astrocytoma diagnosed with and without histologic confirmation Variable Age* Younger than 1 year 1–4 years 5–9 years 10–19 years 20–29 years 30–39 years 40–49 years 50 years and older Gender, female Race Caucasian Black Others or unknown Laterality* Right, single primary Left, single primary Bilateral, primary Not a paired site or midline† Unknown or unspecified Treatment‡ Tumor-directed surgery Radiation (with or without surgery) Chemotherapy Outcome* Alive or dead of other causes Death attributable to neoplasm Not applicable when another primary tumor known Follow-up (months)§ Median (interquartile range) Histologic Confirmation 169 (54.0%) 19 55 32 26 10 6 11 10 88 Radiologic Diagnosed Without Biopsy 144 (46.0%) (11.2) (32.5) (18.9) (15.4) (5.9) (3.6) (6.5) (5.9) (52.1) 4 (2.8) 62 (43.1) 49 (34.0) 22 (15.3) 2 (1.4) 3 (2.1) 0 (0.0) 2 (1.4) 64 (44.4) 137 (81.1) 13 (7.7) 19 (11.3) 119 (82.6) 12 (8.3) 13 (9.1) 51 39 11 58 10 (30.2) (23.1) (6.5) (34.3) (5.9) 126 (74.6) 25 (14.8) 67 (39.6) 151 (89.3) 12 (7.1) 6 (3.6) 113.5 (49.5, 179) 48 44 40 3 9 (33.3) (30.6) (27.8) (2.1) (6.3) 0 (0)* 2 (1.4)* 19 (13.2)* 129 (89.6) 1 (0.7) 14 (9.7) 42.5 (19,77) *Statistically significant ,0.01, chi-square test. † Assumed chiasm when coded as “not a paired site” or as midline. ‡ Tumor-directed surgery, radiation, and chemotherapy recorded as unknown were combined with “not performed.” § Statistically significant P , 0.01, Mann–Whitney U test. diagnosis were statistically significant associations (see Supplemental Digital Content, Table 7A, http://links.lww.com/ WNO/A605). There were no statistically significant univariate associations for PA (see Supplemental Digital Content, Table 8A, http://links.lww.com/WNO/A606). The adjusted logistics regression model for optic nerve glioma found positive histologic biopsy (right eye 4.66; 99% CI 1.16–18.72) and age older than 20 years (OR 10.79; 99% CI: 2.64– 44.22) significant associations (see Supplemental Digital Content, Table 7B, http://links.lww.com/WNO/A605). No variables reached statistical significance (P , 0.01) for PA in the adjusted model although diagnostic (biopsy) confirmation is trending toward significance (see Supplemental Digital Content, Table 8B, http://links.lww.com/WNO/A606). DISCUSSION The results of our study support the idea that optic nerve gliomas coded in SEER as PA (9421/3) and glioma (9380/3) Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 are mostly low-grade neoplasms that occur in persons younger than 20 years of age. The similarities in age profiles and overall clinical outcomes also indicate that these 2 diagnostic terms were used interchangeably in SEER. Why individual cases were placed in one category over another is unclear. The fact that a specific histopathology diagnosis of PA was made in the absence of a biopsy in 46% of patients is problematic. The noncommittal and presumptive diagnosis of “glioma” would be appropriate for a radiologic diagnosis, but without tissue, neuroimaging cannot exclude other types of low-grade astrocytomas (e.g., low-grade diffuse glioma, pilomyxoid astrocytoma, etc.) (3–5,7,8). Given past evidence that the behavior in a subset of optic nerve gliomas cannot be predicted by radiologic interpretation alone, tissue-specific histopathologic diagnoses should have been avoided in the absence of a biopsy. To place this problem into perspective, neoplastic astrocytes exist along a morphologic spectrum and are characterized by the expression of glial acid fibrillary protein. 465 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 4. Glioma diagnosed with and without histologic confirmation Variable Age Younger than 1 year 1–4 years 5–9 years 10–19 years 20–29 years 30–39 years 40–49 years 50 years and older Gender, female Race Caucasian Black Others or unknown Laterality Right, single primary Left, single primary Bilateral, primary Not a paired site or midline* Unknown or unspecified Treatment† Tumor-directed surgery Radiation (with or without surgery) Chemotherapy Outcome‡ Alive or dead of other causes Death attributable to neoplasm Not applicable when another primary tumor known Follow-up (months) Median (interquartile range) Histologic Confirmation 95 (13.2%) 3 44 17 9 6 2 7 7 51 (3.2) (46.3) (17.9) (9.5) (6.3) (2.1) (7.4) (7.4) (53.7) Radiologic Diagnosis Without Biopsy 625 (86.8%) 42 258 148 90 21 22 26 18 332 (6.7) (41.3) (23.7) (14.4) (3.4) (3.5) (4.2) (2.9) (53.1) 79 (83.2) 10 (10.5) 6 (6.3) 522 (83.5) 49 (7.8) 54 (8.6) 26 20 12 31 6 171 156 114 135 49 (27.4) (21.1) (12.6) (32.7) (6.3) (27.4) (25.0) (18.2) (21.6) (7.8) 26 (27.4)‡ 16 (16.9)‡ 35 (36.8)‡ 2 (0.3) 38 (6.1) 158 (25.3) 84 (88.4) 7 (7.4) 4 (4.2) 589 (94.2) 7 (1.1) 28 (4.5) 98 (46, 177) 92 (44, 149) *Assumed chiasm when coded as not a paired site or midline. † Single patient may have more than 1 therapy. ‡ Statistically significant difference at P , 0.01, chi square test. Collectively referred to as gliomas, the WHO classifies astrocytomas based on microscopic similarities with putative cells of origin (7,8). Although histopathology is the standard for classification of low-grade gliomas, there is high intraobserver and interobserver variabilities among well-trained pathologists and grading often does not predict the outcome (13–15). The limitations of morphologic classification have led to efforts to enhance prognostic capabilities through the identification of molecular and genetic alterations (15–17). PA, the most common glioma of optic nerve in childhood, is characterized by variable densities of delicate bipolar cells, Rosenthal fibers, and microcysts. It is a low-grade, benign tumor (WHO Grade 1) associated with mutations in the mitogen-activated protein kinase pathway (15,17). Unlike diffuse fibrillary astrocytomas (low-grade [WHO Grade II]), PAs rarely undergo malignant transformation to anaplastic astrocytoma (WHO Grade III) or glioblastoma (WHO Grade IV) (8). PA is also the principal neoplasm of the 466 central nervous system in persons with neurofibromatosis Type 1 (NF1) and whose pathogenesis is linked to the overactivity of the RAS and the MAPK pathways (7,15). In the SEER data set, both PAs and gliomas after biopsy were managed differently than their nonbiopsied counterparts and had different outcomes. Why? The results of adjusted regression analysis that showed age older than 20 years in persons with glioma was associated with tumorrelated death (see Supplemental Digital Content, Table 7, http://links.lww.com/WNO/A605) does little to explain why the method of diagnosis was linked to outcome. The reasons for differences in outcome related to tumor coding (glioma vs PA) and method of diagnosis are likely due to factors such as bias created by anatomic misclassification and failure to collect key variables truly related to the behavior of the tumors, such as molecular and genetic factors (e.g., NF1 status, etc.) or extent of disease (spread to hypothalamus, etc.). Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 5. Pilocytic astrocytoma: birth–19 years old and 20 years and older Variable Birth—19 Years 269 (85.9%) Gender, female Race Caucasian Black Others or unknown Laterality* Right, single primary Left, single primary Bilateral, primary Not a paired site or midline† Unknown or unspecified Treatment‡ Tumor-directed surgery Radiation (with or without surgery) Chemotherapy Outcome Alive or dead of other causes Death attributable to neoplasm Not applicable when another primary tumor known Follow-up (months)§ Median (interquartile range) 20 Years and Older 44 (14.1%) 127 (47.2) 25 (56.8) 218 (81.0) 22 (8.2) 29 (10.8) 38 (86.4) 3 (6.8) 3 (6.8) 79 78 49 46 17 (29.4) (29.0) (18.2) (17.1) (6.3) 20 5 2 15 2 (45.5) (11.4) (4.5) (34.1) (4.6) 102 (37.9) 10 (3.7) 83 (30.9) 24 (54.5) 17 (38.6)* 3 (6.8)* 243 (90.3) 10 (3.7) 16 (5.9) 37 (84.1) 3 (6.8) 4 (9.1) 67 (28,127) 102 (30,202) *Statistically significant P # 0.01, chi square test. † Combined with midline when coded as “not a paired site.” ‡ Single patient may have more than 1 therapy. § Trending toward statistical significance P = 0.03, Mann–Whitney U test. Approximately 15% of PAs and gliomas in SEER occurred in persons older than 19 years, which is slightly higher than the 10% noted by Dutton in his review of gliomas of the anterior visual pathway (1). The study by Dutton also recorded a separate category of so-called adult gliomas (ranging in age from 6 years to 79 years) that were clinically and histologically malignant. The 41 patients in the review by Dutton with so-called adult gliomas had uniformly poor outcomes and were distinct from the more common indolent optic nerve gliomas of childhood (1). Although the percent of attributable deaths for PA and glioma in SEER was higher in persons 20 years and older (6.8% for PA and 9.2% for glioma), these proportions are far less than that anticipated for true malignant gliomas of the optic nerve (1,18). Numerous investigators have attempted to better predict clinical behavior and the propensity for local invasion of low-grade gliomas, a problem which is challenging for the anterior visual pathway because of the morbidity associated with biopsy and the limitations of histologic classification (18–22). Although most cases of low-grade optic nerve glioma display characteristic features on MRI, there is yet no definitive means of predicting aggressive growth among them (3–5,23–33). Since submission of our article, another SEER study of optic nerve glioma has been published using data from 1984 to 2016 (34). Their findings generally parallel ours, but they did not test the hypothesis that method of diagDihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 nosis may influence outcome. Relevant to our study, however, only 28% of the 1,004 cases they reported had tissue diagnoses; yet 94.9% (953) were considered low-grade astrocytomas (i.e., Grade I and II, WHO criteria) (34). To the best of our knowledge, there is no evidence that a Grade I (PA) and Grade II astrocytoma can be reliably distinguished from one another by neuroimaging. The limitations of the SEER data set have been described elsewhere and importantly involve missing data particularly variables that influence prognosis (35–38). Several problems are specific to the study of tumors of the optic nerve, the most relevant of which is the failure of ICD-O topography code for the optic nerve (C72.3) to distinguish true optic nerve from chiasm and optic tract. A left side astrocytoma confined to the optic tract, for example, would be coded as the left optic nerve. Inaccurate anatomical classification results in a differential bias in the collection of information that can undermine the validity of studies. The influence of this bias can be understood by the fact that low-grade gliomas posterior to the chiasm are generally more likely to have poor outcomes than an isolated optic nerve glioma (39). SEER also lacks critical information on variables of prognostic value including molecular/genetic markers and NF1 status. These deficiencies will become more critical now that targeted therapies such as MEK inhibitors are being used in low-grade gliomas (39). 467 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 6. Glioma: birth to 19 years old and 20 years and older Variable Birth—19 Years 611 (84.9%) Gender, female Race Caucasian Black Others and unknown Laterality* Right, single primary Left, single primary Bilateral, primary Not a paired site or midline† Unknown or unspecified Treatment*‡ Tumor-directed surgery Radiation (with or without surgery) Chemotherapy Outcome Alive or dead of other causes Death attributable to neoplasm Not applicable when another primary tumor known Follow-up (months)§ Median (interquartile range) 20 Years and Older 109 (15.1%) 316 (51.7) 67 (61.5) 514 (84.1) 45 (7.4) 52 (8.6) 87 (79.8) 14 (12.8) 8 (7.3) 153 147 120 148 43 44 29 6 18 12 (25.0) (24.1) (19.6) (24.2) (7.0) (40.4) (26.6) (5.5) (16.5) (11.0) 17 (2.8) 13 (2.2) 181 (29.6) 11 (10.1) 41 (37.6) 12 (11.0) 585 (95.7) 4 (0.7) 21 (3.4) 88 (80.7) 10 (9.2) 11 (10.1) 96 (49, 158) 77 (27, 130) *Statistically significant at P # 0.01, chi square test. † Assumed chiasm when coded as not a paired site or midline. ‡ Single patient may have more than 1 therapy. § Statistically significant at P = 0.01, Mann–Whitney U test. Failure to identify patients with NF1 in SEER could also contribute to an artifactitious loss of patients in follow-up (outcome) because persons with neurofibromatosis are prone to have multiple primary tumors. A second tumor withdraws patients in the registry from further follow-up and could possibly explain the imbalance in follow-up found among groups in this study. CONCLUSIONS The SEER data set on optic nerve glioma of childhood reveals that from 1975 through 2017 about two-thirds were classified as glioma and a third as PA. Nearly half of PAs diagnosed by imaging alone were given a tissue-specific histopathologic diagnosis rather than the nonspecific designation of glioma. Overall survival was high in both groups, but tumor attributable death was greater among PA than glioma (4.2% vs 1.9%) and higher among those undergoing biopsy within each diagnostic group. The data suggest that accurate anatomical location and clinical variables relevant to clinical decision-making were not captured by the tumor registry. The data indicate there is a subset of neoplasms that have a more aggressive clinical course than typical PA. The possibility that histopathologic variants with more aggressive behavior (e.g., low-grade diffuse glioma, pilomyxoid astrocytoma, etc.) are being missed cannot be tested because a majority of cases were diagnosed by radiologic criteria only (18–21,39). Tumor registries need to consider modifying the 468 information they collect on tumors of the anterior visual pathway including precise anatomical location, extent of tumor involvement, and key molecular/genetic markers known to be linked to prognosis. Tumor registries need to address the practice of coding for specific histopathologic diagnoses when biopsies have not been performed. STATEMENT OF AUTHORSHIP Conception and design: F. Dihowm and C. E. Margo. Acquisition of data: F. Dihowm and L. A. Alvarado. Analysis and interpretation of data: F. Dihowm, C. E. Margo, and L. A. Alvarado. Drafting the manuscript: C. E. Margo. Revising the manuscript for intellectual content: C. E. Margo, F. Dihowm, and L. A. Alvarado. Final approval of the completed manuscript: C. E. Margo, F. Dihowm, and L. A. Alvarado. REFERENCES 1. Dutton JJ. Gliomas of the anterior visual pathway. Surv Ophthalmol. 1994;38:427–452. 2. Pepin SM, Lessell S. Anterior visual pathway gliomas: the last 30 years. Semin Ophthalmol. 2006;21:117–124. 3. Fried I, Tabori U, Tihan T, Reginald A, Bouffet E. Optic pathway gliomas: a review. CNS Oncol. 2013;2:143–159. 4. Rassol N, Odel JG, Kazim M. Optic pathway glioma of childhood. Curr Opin Ophthalmol. 2017;28:289–295. 5. Nair AG, Pathak RS, Iyer VR, Gandhi RA. Optic nerve glioma: an update. Int Ophthalmol. 2014;34:999–1005. 6. Farazhaghi MK, Katowitz WR, Avery RA. Current treatment of optic nerve glioma. Curr Opin Ophthalmol. 2019;30:356–363. 7. Louis DN, Ohgaki H, Wiestler OD, Cavenee WK. WHO Classification of Tumours of the Central Nervous System. Lyon, France: International Agency for Research on Cancer, 2016. Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution 8. Ironside JW, Moss TH, Louis DN, Lowe JS, Weller RO. Diagnostic Pathology of Nervous System Tumours. London, United Kingdom: Churchill Livingston, 2002. 9. SEER Inquiry System. Available at: https://seer.cancer.gov/ seerinquiry/index.php?page=view&id=20200079&type=q. Accessed November 5, 2021. 10. Peckham-Gregory EC, Montenegro RE, Stevenson DA, Viskochil DH, Scheurer ME, Lupo PJ, Schiffman JD. Evaluation of racial disparities in pediatric optic pathway glioma incidence: results from the Surveillance, Epidemiology, and End Results Program, 2000–2014. Cancer Epi. 2018;54:90–94. 11. Mishra MV, Andrews DW, Glass J, Evans JJ, Dicker AP, Shen X, Lawrence YR. Characterization and outcomes of optic nerve gliomas: a population-based analysis. J Neurooncol. 2012;107:591–597. 12. Available at: https://www.seer.cancer.gov.mmm.seer.cancer. gov/about/overview.html. Accessed January 11, 2020– January 12, 2020. 13. Coons SW, Johnson PC, Scheithauer BW, Yates AJ, Pearl DK. Improving diagnostic accuracy and interobserver concordance in the classification and grading of primary gliomas. Cancer. 1997;79:1381–1393. 14. van den Bent MJ. Interobserver variation of the histopathological diagnosis in clinical trials on glioma. A clinician’s perspective. Acta Neuropathol. 2010;120:297–304. 15. Collins VP, Jones DT, Giannini C. Pilocytic astrocytoma: pathology, molecular mechanisms and markers. Acta Neuropathol. 2015;129:775–788. 16. The Cancer Genome Atlas Research Network. Comprehensive, integrative genomic analysis of diffuse lowergrade gliomas. N Engl J Med. 2015;372:2481–2498. 17. Gutmann DH, Parada LF, Silva AJ, Ratner N. Neurofibromatosis type 1: modeling CNS dysfunction. J Neurosci. 2012;32:14087–14093. 18. Traber G, Pangalu A, Neumann M, Costa J, Weller M, HunaBaron R, Landau K. Malignant optic glioma—the spectrum of disease in a case series. Graefes Arch Clin Exp Ophthalmol. 1015;253:1187–1194. 19. Matyja E, Grajkowska W, Stepien K, Naganska E. Heterogeneity of histopathological presentation of pilocytic astrocytoma—diagnostic pitfalls. Folia Neuropathologies. ;54:197–211. 20. Gaijar A, Bhargava R, Jenkins JJ, Heideman R, Sanford RA, Langston JW, Walter AW, Kuttesch JF, Muhlbauer M, Kun LE. Low-grade astrocytoma with neuraxis dissemination at diagnosis. J Neurosurg. 1995;83:67–71. 21. Perilongo G, Carollo C, Salviati L, Murgia A, Pillon M, Basso G, Gardiman M, Laverida A. Diencephalic syndrome and disseminated juvenile pilocytic astrocytoma of the hypothalamic-optic chiasm region. Cancer. 1997;80:142–146. 22. Pollack IF, Hurtt M, Pang D, Albright AL. Dissemination of low grade intracranial astrocytomas in children. Cancer. 1994;73:2869–2878. 23. Kornreich L, Blaser S, Schwarz M, Shuper A, Vishne TH, Cohen IJ, Faingold R, Michovitz S, Koplewitz B, Horev G. Optic pathway glioma: correlation of imaging findings with the presence of neurofibromatosis. AJNR Am J Neuroradiol. 2001;22:1963–1969. Dihowm et al: J Neuro-Ophthalmol 2022; 42: 462-469 24. Listernick R, Ferner RE, Liu GT, Gutmann DH. Optic pathway gliomas in neurofibromatosis-1: controversies and recommendations. Ann Neurol. 2007;61:189–198. 25. Becker M, Masterson K, Delavelle J, Viallon M, Vargas MI, Becker CD. Imaging of the optic nerve. Eur J Radiol. 2010;74:299–313. 26. Weber AL, Caruso P, Sabates NR. The optic nerve: radiologic, clinical, and pathologic evaluation. Neuroimaging Clin N Am. 2005;15:175–201. 27. Jahraus CD, Tarbell NJ. Optic pathway gliomas. Pediatr Blood Cancer. 2006;46:586–596. 28. Binning MJ, Liu JK, Kestle JR, Brockmeyer DL, Walker ML. Optic pathway gliomas: a review. Neurosurg Focus. 2007;23:E2. 29. Liu GT. Optic gliomas of the anterior visual pathway. Curr Opin Ophthalmol. 2006;17:427–431. 30. Sylvester CL, Drohan LA, Sergott RC. Optic-nerve gliomas, chiasmal gliomas and neurofibromatosis type 1. Curr Opin Ophthalmol. 2006;17:7–11. 31. Pascual-Castroviejo I, Pascual-Pascual SI, Velazquez-Fragua R, Viaño J, García-Segura JM, Botella MP. Neurofibromatosis type 1 and optic pathway gliomas. A series of 80 patients. Rev Neurol. 2008;49:530–5306. 32. Nicolin G, Parkin P, Mabbott D, Hargrave D, Bartels U, Tabori U, Rutka J, Buncic JR, Bouffet E. Natural history and outcome of optic pathway gliomas in children. Pediatr Blood Cancer. 2009;53:1231–1237. 33. Astrup J. Natural history and clinical management of optic pathway glioma. Br J Neurosurg. 2003;17:327–335. 34. Zhang M, Chen T, Zhong Y. Demographic and prognostic factors of optic nerve astrocytoma: a retrospective study of surveillance, epidemiology, and end results (SEER). BMC Cancer. 2021;21:976–984. 35. Yu JB, Gross CP, Wilson LD, Smith BD. NCI SEER public-use data: applications and limitations in oncology research. Oncology (Williston Park). 2009;23:288–295. 36. Shaikh WR, Weinstock MA, Halpern AC, Oliveria SA, Geller AC, Dusza SW. The characterization and potential impact of melanoma cases with unknown thickness in the United States’ Surveillance, Epidemiology, and End Results Program, 1989– 2008. Cancer Epidemiol. 2013;37:64–70. 37. Kim HM, Goodman M, Kim BI, Ward KC. Frequency and determinants of missing data in clinical and prognostic variables recently added to SEER. J Registry Manag. 2011;38:120–131. 38. Ostrom QT, Gittleman H, Kruchko C, Louis DN, Brat DJ, Gilbert MR, Petkov VI, Barnholtz-Sloan JS. Completeness of required site-specific factors for brain and CNS tumors in the Surveillance, Epidemiology and End Results (SEER) 18 database (2004–2012, varying). J Neurooncol. 2016;130:31–42. 39. Colin C, Padovani L, Chappé C, Mercurio S, Scavarda D, Loundou A, Frassineti F, André N, Bouvier C, Korshunov A, Lena G, Figarella-Branger D. Outcome analysis of childhood pilocytic astrocytomas: a retrospective study of 148 cases at a single institution. Neuropathol Appl Neurobiol. 2013;39:693–705. 469 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited.