Chapter

ASTROCYTIC TUMORS I

One of the frequent pitfalls in the prognostic diagnosis of gliomas is encountered in the most classical and easy tumor: diffuse astrocytoma. Classically, the diffuse astrocytoma, subdivided into fibrillary, protoplasmic and gemistocytic variant, does not show problems of recognition. The three variants when fully expressed do not leave doubts as to their identification. Difficulties arise when the surgical samples are too small, as in stereotactic biopsies, or they have not been removed from the full tumor, so that they do not contain cell distribution patterns which make the tumor recognizable. If the sample is taken from the tumor periphery, all normal cells of the nervous tissue will be present and it is common knowledge that one cell of astrocytoma, individually, cannot be distinguished from a normal or reactive astrocyte.

Many diagnostic difficulties can be encountered in anaplastic astrocytomas and glioblastomas, especially when the surgical sample is taken purposely from a non-necrotic area or there is a sampling error. Stereotactic biopsy is the main procedure for a pre-surgical histological diagnosis with low risk, accuracy and with minimally invasive nature. However, it has limitations and not infrequently it is useless up to the point that it has even been regarded as unnecessary (Vaquero et al., 2000; Jackson et al., 2001).

1. DIFFERENTIATION OF TUMOR VS NON-TUMOR TISSUE

Very frequently the small sample is removed from the very periphery of the tumor or from a lesion that is not a tumor, for example ischemic or demyelinating areas (Figure 1). This aspect can be dominated by apparently normal cells and the first question is to establish whether their number has increased, they are glial or neuronal or both, and whether at least some of them are neoplastic.

Figure 1. Hypodense ischemic insular lesion, MRI, T1. From the Neuroradiology Unit, Dpt Neuroscience, University of Turin

The increase in the number of cells cannot always be easily assessed, especially when it does not exceed 30%. Sometimes the number of nuclei seems to be normal, but their volume is increased (Figure 3A, B). Normal neurons are easily recognizable, once the existence of a neuronal tumor has been excluded. Inflammatory cells can be recognized by the small and dark nucleus and by their crowding around the vessels in addition to being scattered thoughout the tissue. Microglia cells are also easily recognizable by the irregular form of the nucleus. Specific immunohistochemical methods will be of help, such as those employing CD3, CD20, CD68 (Figure 2A, B).

In the cerebral cortex, the number of peri-neuronal satellites, either oligodendroglial or astroglial, may be increased and, letting aside frank aspects of satellitosis that may not be so easily attributed to astrocytomas or oligodendrogliomas, quite often it is even difficult to assess (Figure 4A).

Microglial Expansion Cd68

Figure 2. A. Microglia cells in the cortex, CD68, DAB, x 400; B. Inflammatory cells in the cortex, H&E, x 400

Figure 3. Stereotactic biopsies from a glioma periphery. A. Likely increase of nuclei in the white matter, H&E, x 400; B. Increase of the nuclear volumes, H&E, x 400

Figure 4 . A. Infiltrated cortex. Peri-neuronal satellitosis, H&E, x 400; B. Reactive astrocytes in the white matter,

Figure 4 . A. Infiltrated cortex. Peri-neuronal satellitosis, H&E, x 400; B. Reactive astrocytes in the white matter,

Once the cells which have increased in number have been recognized as astroglial on the basis of their nucleo-cytoplasmic appearance, they must be identified as tumor or reactive astrocytes. Reactive astrocytes are identifiable for their clear and large nuclei and the eosinophilic cytoplasm and for its long and thick GFAP-positive processes (Figure 4B). When the gliosis is not recent and it has not been renewed for some time, all the astrocytes show the same aspect and are found at regular intercellular distances. However, they are frequently generated at different times, as happens in infiltrating tumors for example, so that not all of them are at the same stage of reaction and the recently formed ones could not be recognized as reactive. The pre-existing GFAP-positive glial network may still be visible, and it may create some difficulties in the diagnosis when small reactive astrocytes are contemporarily present, simulating a low cell density astrocytoma (Figure 5A, B). Positive staining for MIB-1 may help, taking into account the fact that also reactive astrocytes may show positive nuclei, with a mean LI slightly lower than that of astrocytomas, but with more or less similar LI ranges (Wessels et al., 2001). Also positive staining for p53 may be of some help in favor of an astrocytoma when it is positive.

Even after recognizing reactive astrocytes as the major cell component in the section, the occurrence of a tumor cannot be discounted with certainty, because a massive peripheral gliosis my mask scattered neoplastic astrocytes. The examination of MRI is mandatory and it can maybe orientate the pathologist according to the type of lesion, with or without contrast enhancement or simply given by a small hypo-intense spot. The progression over time of the lesion or the appearance of contrast enhancement will confirm the neoplastic nature.

Studies with in situ hybridization showing chromosomal aberrations in all interphase cells have been instructive. Alterations of chromosome 1, 7 and 10, characteristic of low-grade astrocytomas (Rosso et al., 1997; Perry et al., 1997; Hopman and Ramaekers, 1998) have been found in 80% of astrocytomatous areas, but never in reactive astrocytes (Wessels et al., 2001). In the last series, three out of four cases with inconclusive diagnosis between glial reaction and low-grade astrocytoma, showed aneusomies and underwent rapid progression indicating that they were unrecognized high-grade tumors. It is important to note that no single genetic aberration has been shown to be responsible for tumor progression. Coupling microdissection techniques with PCR reactions for individual polymorphic microsatellites situated at 7 genomic regions (1p, 3p, 5q, 9p, 1oq, 17p, 19q), no allelic loss occurred in reactive gliosis against at least one loss in gliomas of every grade. In 73% of cases of uncertain attribution, the occurrence or not of LOH allowed a correct prediction (Finkelstein et al., 2004). This observation is very important, because it demonstrates that molecular genetics can be applied to minute formalin-fixed and paraffin-embedded specimens for a molecular analysis in everyday pathology practice.

It is of paramount importance to consider that, even when quite certain, the diagnosis of gliosis does not exclude the possibility of an adjacent tumor. In this case, the careful examination of clinical and imaging data is mandatory and in some cases, if the latter are in contrast with the pathology, another biopsy can be prompted.

It must be recalled that sometimes gliosis corresponds to a demyelinating disease or to an infarction; and the differential diagnosis must take into account the fact that the occurrence of macrophages, even if constant in these lesions, is not their exclusive characteristic, being frequent also in malignant tumors (Figures 6A, B; 7A, B).

Figure 5. Glioma periphery. A. Pre-existing GFAP-positive reticulum and small reactive astrocytes, DAB x 400; B. Early reactive astrocytes, H&E, x 400

Figure 6. Demyelinating area. A. Increased number of nuclei with reactive astrocytes on the right, H&E, x 400; B. Myelin loss on the right, Luxol Fast B, x 400

Figure 7. A. Reactive astrocytes in the demyelinating area, GFAP, DAB, x 400; B. Macrophages in glioblastoma, CD68, DAB, x 400

Figure 7. A. Reactive astrocytes in the demyelinating area, GFAP, DAB, x 400; B. Macrophages in glioblastoma, CD68, DAB, x 400

2 . DIFFUSE ASTROCYTOMA VS OLIGODENDROGLIOMA

Frequently, the differential diagnosis using a small fragment from the periphery of a tumor, that at MRI appears as a circumscribed hypo-intense lesion (Figure 8), must be carried out between a diffuse astrocytoma and the diffuse growth of an oligodendroglioma.

Figure 9. Reactive astrocytes with oligodendrocytes not increased in number, but with cytoplasmic halos.

Figure 9. Reactive astrocytes with oligodendrocytes not increased in number, but with cytoplasmic halos.

Classically, the distinction should be based on the aspect of the nuclei, the "chicken wire" distribution of small vessels (Figure 10), the occurrence of "honeycomb" appearance indicating oligodendroglioma (Figure 9) and on larger and more vesiculous nuclei and GFAP-positive staining indicating astrocytoma, but very rarely are these aspects so clear-cut (Figures 11, 12). In particular, nuclei are small and round, with a thick membrane and a central small nucleolus in oligodendrogliomas (Figure 15) and larger and vesicular with a very small nucleolus in astrocytomas. However, confusion can originate from the possibility that tumor cells are GFAP negative in astrocytoma and positive in oligodendroglioma, like minigemistocytes and GFOC (gliofibrillary oligodendrocytes) (Figures 6-13, 14) (Herpers and Budka, 1984; Schiffer, 1997).

Figure 10. Small vessels of oligodendrogliomatous type, H&E, x 400

When tumor proliferation is in the white matter and small reactive astrocytes are present, the pre-existent GFAP-positive glial network being still visible, normal oligodendroglial nuclei could be confused with tumor oligodendrocytes and this could lead to the erroneous diagnosis of oligodendroglioma or of oligoastrocytoma (Figures 11-13). The opposite is also possible when small reactive astrocytes are overestimated, especially when associated with a GFAP-positive glial network, and oligodendroglial nuclei are considered as belonging to normal oligodendroglia of the white matter (Figures 11, 14). Sometimes it is difficult to establish whether GFAP-positive, astrocyte-like cells are minigemistocytes or tumor or reactive astrocytes (Figure 16).

Figure 11. Reactive astrocytes and normal or tumor oligodendroglial nuclei, H&E, x 400
Figure 12. Field uncertain between diffuse astrocytoma and oligodendroglioma, H&E, x 400
Diffuse Astrocytoma

Figure 13. Field close to that of Figure 9: the tumor appears to be an oligodendroglioma with GFAP reactive astrocytes, DAB, x 400

Figure 13. Field close to that of Figure 9: the tumor appears to be an oligodendroglioma with GFAP reactive astrocytes, DAB, x 400

Sophisticated cytometric procedures on Feulgen-stained sections could help in distinguishing astrocytic from oligodendrocytic nuclei (Deckaestecker et al., 1997), but they are not easily applicable, especially when the diagnosis must be given without delay. The old procedures with acetic carmin could be of help and are more practicable (Schiffer and Fabiani, 1971), but they are today obsolete. Since there is no marker for the recognition of tumor oligodendrocytes in paraffin sections, the distinction between normal and tumor oligodendrocytes may be important and this can be achieved by the immunohistochemical demonstration of Cyclin D1 which is positive in normal oligodendrocytes and not in most tumor oligodendrocytes, unless these are cycling cells (Bosone et al., 2001; Fiano et al., 2003) (See chapter VI).

In some cases, not only for the reduced dimensions of the specimen, but also because of the diffuse hypo-intensity at MRI that shows nothing characteristic for one tumor or the other, the distinction is really difficult and a diagnostic compromise is chosen with the diagnosis of oligoastrocytoma. From the point of view of the therapeutic decisions to be taken by the clinician, this uncertainty has no treatment-affecting consequence. It becomes a real problem when possible anaplastic features occur and an interpretation must be given, for example, to the MIB-1 LI, for establishing the malignancy grade. MIB-1 LI is <4% in astrocytomas, whereas in oligodendrogliomas it can reach 15% in grade II tumors (Schiffer et al., 1997). With a MIB-1 LI of 10%, for example, the grade will be the IIIrd if the

Figure 14. Reactive astrocytes and oligodendreoglial nuclei, H&E, x 400

diagnosis is that of astrocytoma and the IInd if it is that of oligodendroglioma. The confusion may have grave therapeutic consequences with irradiation and chemotherapy of an oligodendroglioma grade II as if it were an anaplastic astrocytoma and vice versa. Another serious consequence could be that of expecting an outcome of anaplastic astrocytoma from a tumor that is a grade II oligodendroglioma. This can be a source of mistakes when survival is used as a criterion for evaluating the efficacy of therapies. This matter will be further discussed with consideration of the oligodendrogliomas.

The possibility to have some help from non histological procedures is time-consuming and costly. In research directed to the identification of molecular markers of astrocytomas and oligodendrogliomas it has been shown that the frequency of galactosyltransferase (CGT) transcripts were twice as frequent in oligodendrogliomas than in astrocytomas and that GA1 (asialo GM1) was most frequent in oligodendrogliomas associated with the absence of paragloboside (Popko et al., 2002). The feasibility of this test however, must still be checked and its usefulness in individual cases seems low. More important can be the procedures for the identification of 1p and 19q losses, provided that the time required is reasonable. Of these more will be said later on.

3. MALIGNANCY GRADE IN RELATION TO PROGNOSIS AND THERAPIES

Basically, the starting point for every neuropathologist facing the problem of histological diagnosis of astrocytic tumors in small fragments is that most diffuse

Figure 15. Increased number of frankly oligodendroglial nuclei, H&E, x 400

astrocytomas transform in time into glioblastomas, through the occurrence of anaplasia, and that this transformation may already be in progress at the moment of the diagnosis, even when either phenotypic or neuro-imaging alterations are not yet present (Figure 17). The occurrence of non-detectable transformed foci in a diffuse astrocytoma grade II has been considered among the rationales for radiotherapy of these tumors, with the goal of sterilizing them from malignant cells. Many studies have been dedicated to the detection, by every means and recently especially by molecular genetics, of prognostic factors when the phenotype is not or not yet informative.

It is common knowledge that >60% diffuse astrocytomas show TP53 mutations which do not increase with tumor progression. Not every cell contains mutations and with the increase of malignancy grade there can be a clonal expansion (Sidransky et al., 1992). Astrocytoma grade II cannot be further characterized from the molecular genetics point of view. On the contrary, in anaplastic astrocytoma, alterations of pRb pathway are present and can be demonstrated also in histological sections, even though none of them is characteristic of the tumor. They are interchangeable. The cell cycle can be deregulated by the alteration of one or more of the oncogenes/ proteins involved.

None of the four criteria for the recognition of malignancy, i.e. nuclear pleomorphism, mitoses, circumscribed necroses and microvascular proliferations, is present in diffuse astrocytomas. In these cases, MRI shows circumscribed areas

Figure 16. GFAP- positive cells, uncertain between GFOC and tumor or reactive astrocytes, DAB, x 400

hypo-intense in T1 and hyper-intense in T2 weighted images; but when it shows non homogeneous aspects and, especially, contrast enhancement, most probably the suspicion of a sampling error is legitimate, because the histological picture should be of greater malignancy. This confirms incidentally the necessity for the neuropathologist to receive at least some education in neurological clinics and neuro-radiology in order to be able to manage the neuro-imaging information. The possibility of being disavowed by the follow-up of the patient in this way can be reduced, but not completely abolished, because in some cases, in spite of the homogeneous iso- or hypo-intense aspect at MRI and a quiescent histological aspect, the outcome will be that of a malignant tumor, including contrast enhancement appearance in the course of time. The opposite situation is also possible i.e. that a quiescent MRI aspect disguises the occurrence of malignancy signs, maybe limited to nuclear pleomorphism and mitoses. Therefore, at the moment of diagnosis, anaplasia could already have been in progress or impending, but not yet phenotypically expressed, or it may develop later on (Figure 17).

In the differential diagnosis between astrocytoma grades II and III, the greatest importance is attributed to nuclear pleomorphism and mitoses (Figure 19A, B and Figure 20), because circumscribed necroses and microvascular proliferations indicate glioblastoma or grade IV (Figure 18A, B).

In the binary system (Daumas-Duport et al., 1988) applied to astrocytic tumors, nuclear pleomorphism or mitoses alone indicate grade II, whereas together they

Figure 17. Hypodense temporal lesion, with histological characteristics of Grade III astrocytoma, MRI, T1. From the Neuroradiology Unit, Dpt Neuroscience, University of Turin

indicate grade III. In this regard, it has been discussed whether one single mitosis could be enough for recognizing grade III or whether a cut-off of the number of mitoses tolerated in grade II should be used. Almost everybody agrees that one single mitosis cannot indicate grade III, but it is also known that in a sample of astrocytoma grade III it may happen that not more than one single mitosis can be found. The latter observation is of paramount importance, but limited by the questionable representativeness of the sample. In our experience, the cut-off is < 5 mitoses per 10 HPF, but for others it is lower. The significance of a single mitosis found in a sample varies not only according to the extension of the sample, but also and mainly on how many fields must be examined before finding it. It has been observed, for example, that the number of fields at 400 x to be examined for finding one mitosis in astrocytoma grade III is 50, whereas in astrocytoma grade II and IV is 20 (Coons and Pearl, 1998).

The use of mitoses for the evaluation of the malignancy grade is a widespread, very simple and quick system, but many remarks have been addressed as to its reliability (Prayson, 2002): the number of mitoses can decrease for delayed and inadequate fixation and be influenced by the type of staining and the section thickness; different interpretations can be used in defining HPF; mitotic index (MI) can be calculated from the mean value of mitoses counted in all the extension of the section or in fields selected for containing the highest number of mitoses. As has been said before, it is generally acknowledged that it is possible that in an anaplastic astrocytoma not even a single mitosis is found, whereas some may be present in a differentiated astrocytoma.

Figure 18. Glioblastoma. Circumscribed necroses, H&E x 200; B. Microvascular proliferations, H&E, x 200

a"

0 0

Post a comment