Correspondence
Brain Tumors in Children
N Engl J Med 1995; 332:1238-1239May 4, 1995
- Article
To the Editor:
Although the review of childhood brain tumors by Pollack (Dec. 1 issue)1 is excellent, it is based on the assumption that “the initial step in treatment is surgery to establish the histologic diagnosis and to reduce the tumor burden.” The assumption that a histologic diagnosis is correct, predicts biologic behavior, and is a determinant of outcome may not be true for childhood brain tumors.
Histologic features are not necessarily reliably and reproducibly identified by teams of observers2 and are not necessarily assembled reliably into diagnostic categories.3 The objective criteria by which histologic features are used to make diagnoses may be applied inconsistently3 or interpreted by different observers in the same4,5 or separate data bases.4,6,7 The histologic appearance of childhood brain tumors is often heterogeneous, suggesting very different tumors within the same specimen3 — a characteristic that makes efforts to categorize such tumors according to standard World Health Organization (WHO) criteria8 problematic. WHO histologic criteria may be applicable to more than one diagnosis; in one study, the histologic criteria for as many as four disparate tumors were similar.3
Patients enrolled in treatment protocols must be carefully selected so that cohorts are uniform. Currently accepted criteria for establishing histologic diagnoses are relatively arbitrary. The WHO manual8 does not list defining histologic features for each diagnostic class, but simply enumerates “allowable” features. There are few objective definitions of histologic boundaries between and among classes or of the extent to which certain components must be present before a tumor is designated “mixed” or “anaplastic.” If tumors identified by name are not actually uniform and their histologic components are not objectively defined, uniform cohorts cannot possibly be selected for treatment protocols and the resulting data may not have much statistical meaning.
8 ReferencesWilliam D. Brown, M.D.
Floyd H. Gilles, M.D.
Children's Hospital Los Angeles, Los Angeles, CA 90054-07001
Pollack IF . Brain tumors in children. N Engl J Med 1994;331:1500-1507
Full Text | Web of Science | Medline2
Gilles FH, Sobel EL, Leviton A, Tavare CJ, Hedley-Whyte ET, Childhood Brain Tumor Consortium
CrossRef | Web of Science | Medline3
The Childhood Brain Tumor Consortium
Web of Science | Medline4
Schoenberg BS ,Schoenberg DB ,Christine BW ,Gomez MR . The epidemiology of primary intracranial neoplasms of childhood: a population study. Mayo Clin Proc 1976;51:51-56
Web of Science | Medline5
Farwell JR ,Dohrmann GJ ,Flannery JT . Central nervous system tumors in children. Cancer 1977;40:3123-3132
CrossRef | Web of Science | Medline6
Gjerris F ,Klee JG ,Klinken L . Malignancy grade and long-term survival in brain tumours of infancy and childhood. Acta Neurol Scand 1976;53:61-71
CrossRef | Web of Science | Medline7
Gjerris F ,Harmsen A ,Klinken L ,Reske-Nielsen E . Incidence and long-term survival of children with intracranial tumours treated in Denmark 1935-1959. Br J Cancer 1978;38:442-451
CrossRef | Web of Science | Medline8
Kleihaus P, Burger PC, Scheithauer BW. Histological typing of tumors of the central nervous system. 2nd ed. New York: Springer-Verlag, 1993.
Author/Editor Response
Dr. Pollack replies:
To the Editor: Drs. Brown and Gilles raise an important issue that certainly merits further emphasis: the rational selection of a treatment approach for a child with a brain tumor depends on a clear understanding of the identity of the tumor. Their questioning of the importance of establishing a histologic diagnosis is surprising. However, their point is that current histologic criteria for categorizing childhood brain tumors and the diagnostic categories themselves have certain limitations. Although it is usually possible for a neuropathologist to establish a histopathological diagnosis with certainty on the basis of consistent, albeit somewhat arbitrary criteria,1 this is not always the case. In some instances, the diagnosis may be in doubt. The Childhood Brain Tumor Consortium has reported that neuropathologists reviewing the same case may reach different diagnoses in 24 percent of cases.2 However, in general, the areas of disagreement revolve around differences in diagnostic subgrouping (e.g., pilocytic astrocytoma vs. nonpilocytic astrocytoma) rather than differences in major diagnostic categories (e.g., astrocytoma vs. primitive neuroectodermal tumor).
At many institutions, including my own, a three-tiered review process is followed to ensure the correctness of the histopathological diagnosis. An additional level of review is generally required for inclusion in most of the multi-institution studies. First, all brain-tumor specimens are reviewed by a team of at least two neuropathologists. Second, the slides are reviewed in conference by a larger group of neuropathologists before a final diagnosis is established. Third, if the diagnosis is still uncertain, the specimens are forwarded to a number of other neuropathologists for review. Finally, when screening candidates for a cooperative study, one or more neuropathologists generally review the tumor specimens centrally and exclude patients with histopathologically discordant tumors to ensure that the study population consists of patients with tumors that are relatively uniform histologically. In certain cases, the final diagnosis remains controversial, but fortunately this outcome is uncommon. As noted by Brown and Gilles, given the morphologic diversity that may exist between tumors classified as belonging to the same group and even within a particular tumor, it may be important to define more objectively the histologic components of a neoplasm in order to achieve the most reliable and reproducible means of determining the true identity of a tumor. However, even this approach is not infallible, since it is clear that experienced neuropathologists may differ on whether a given histologic feature is present or absent.3
Brown and Gilles are correct in noting that the histologic diagnosis does not always predict biologic behavior. Rather, the histologic appearance of the tumor correlates with biologic behavior in a probabilistic way but does not guarantee that a given tumor will respond in a predictable fashion. For example, low-grade astrocytomas that appear to be almost identical histologically can behave in dramatically different ways. Some remain quiescent after only a subtotal resection with no adjuvant therapy, others are invasive despite multimodal treatment, and rare cases disseminate throughout the neuraxis.4,5 That such biologic diversity may be encompassed by a single histopathological diagnosis indicates the limitations of our current classification schemes. One hopes that it will be possible in the future to integrate information about the histologic features of a tumor with a more detailed understanding of the tumor at the molecular level and of host–tumor interactions in order to refine further the classification and treatment of these neoplasms.
5 ReferencesIan F. Pollack, M.D.
Children's Hospital of Pittsburgh, Pittsburgh, PA 15213-25831
Kleihaus P, Burger PC, Scheithauer BW, et al. Histological typing of tumors of the central nervous system. 2nd ed. New York: Springer-Verlag, 1993.
2
The Childhood Brain Tumor Consortium
Web of Science | Medline3
Gilles FH, Sobel EL, Leviton A, Tavare CJ, Hedley-Whyte ET, Childhood Brain Tumor Consortium
CrossRef | Web of Science | Medline4
Pollack IF ,Hurtt M ,Pang D ,Albright AL . Dissemination of low grade intracranial astrocytomas in children. Cancer 1994;73:2869-2878
CrossRef | Web of Science | Medline5
Hoffman HJ ,Humphreys RP ,Drake JM , et al. Optic pathway/hypothalamic gliomas: a dilemma in management. Pediatr Neurosurg 1993;19:186-195
CrossRef | Web of Science | Medline
