Studying the heterogeneity of brain tumors using medium throughput LOH analysis

The theory of tumor progression by clonal evolution posits that a succession of increasingly malignant clones arise, in turn, driven by the step-wise accumulation of mutations (1). At each point in this process it is thought that one cell in the currently dominant clone acquires a new mutation, allowing it to outcompete its siblings. As a result, its descendants enjoy a higher rate of net growth, making them the new dominant clone whose growth eclipses that of preceding clones, until it too is left behind in the wake of the next mutation. Some direct evidence for this model exists, for example, from analyzing the role of TP53 in brain tumors. In an elegant study of primary and recurrent glioma, it was possible to find the mutation prevalent in recurrent tumors represented in some cells of the primary lesion, suggesting that this molecular defect marked the dominant clone about to emerge (2). In the strictest interpretation of the clonal evolution model, a single clone makes up the bulk of the tumor at a given stage (Fig. 1a). Any observed tumor heterogeneity is considered to be due to the remains of ancestral clones, whose sidelined descendants may persist but do not define the clinical course of the disease. This view fits the data from many studies where tumors are considered to be, to a first approximation, homogeneous entities, and all of a tumor's cells are combined in one sample and analyzed using molecular markers of progression. This interpretation also is reassuring in its implication that effective treatment will result from targeting the dominant clone. The validity of the inherent assumptions that a tumor can be treated as a single entity is an increasingly important concern, particularly as new therapies based on the perceived biological characteristics of an individual patient's tumor, become available. Schematic illustrating the Theory of Clonal Evolution. (a) The dominant clone interpretation suggests that at each mutational event (arrows) a new clone arises, which outcompetes its siblings by virtue of a higher net growth rate. A single clone is thought to contribute the bulk of the tumor and determine its clinical behavior. Heterogeneity of the tumor is thought to be limited and due to the presence of persisting ancestral clones that are numerically and clinically insignificant. (b) In this refinement of the model the persistence of ancestral clones is considered to contribute to a larger degree to the tumor, despite the higher net growth of more progressed clones. Progression is still essentially linear in that the currently most aggressive clone is likely to give rise to the next stage in the tumor's evolution. (c) A further refinement of the theory now allows that clones responsible for tumor progression can arise from numerically inferior ancestral clones. Although what may be termed the dominant clone interpretation of the theory of clonal evolution may very well describe many tumor types, particularly those characterized by extreme changes in growth rate and invasiveness with concomitant well-defined histologic stages, it may not explain all cancers. Here we discuss data from analyses of loss of heterozygosity (LOH), which measures clonally inherited genetic damage associated with the inactivation of tumor suppressor genes (3). Therefore LOH relates directly to cancer progression and also serves as a tool for following clonal patterns. Our data suggest that other views of the theory of clonal evolution besides the dominant clone interpretation may need to be considered in some tumor types. LOH on the short arm of chromosome 1 has recently been found to have prognostic value for patients with gliomas (4). Initially it was found that 1p LOH correlated with treatment response in oligododendroglioma (5), but this observation was extended subsequently to other glioma types, including astrocytoma. In the hands of these investigators, who use 3 CA-repeat polymorphism markers to assess LOH, 1p is said to be lost in its entirety. Accordingly, in the literature, tumors are scored for either the presence or absence of LOH in this region and not on a marker-by-marker basis. LOH incidence is reported to be about 70% for oligodendroglioma. A similar frequency of LOH was found in an initial series of oligodendrogliomas examined in our laboratory. In these studies, paraffin sections that were histologically determined to consist entirely of tumor material were used as a source of tumor DNA, and normal DNA was extracted from peripheral blood, as described previously (6). Following separate polymerase chain reaction (PCR) amplification of CA-repeat polymorphisms, again as described, reactions were pooled and electrophoresed on a CEQ 2000XL capillary electrophoresis device (Beckman Coulter, Inc., Fullerton, CA) according to the manufacturer's instructions. Peak area values generated by the CEQ 2000XL analysis software were used as a measure of allele intensities, and ratios between alleles obtained from normal and tumor DNA were calculated. LOH was said to have occurred when one allele was found to have decreased 40% or more relative to its expected intensity. Five of eight tumors (62.5%) showed LOH of at least one of the three 1p markers used: D1S214, D1S199, and D1S2734. Only the first of these markers is different from those used in other studies: D1S214 was substituted for technical reasons for the nearby D1S508 used by Louis and colleagues (4). However, of the five tumors with loss on 1p, only two showed LOH at all markers that were informative, whereas three showed a mixture of LOH and maintenance. Two of these three were represented by a single biopsy each and showed LOH at D1S199 and maintenance at the other two loci. These data alone suggest that LOH at 1p in olidodendroglioma may not always be due to loss of the entire short arm of chromosome 1. Information regarding the issue of clonality was obtained from the third oligodendrogliomas, which showed a heterogeneous pattern of 1p loss. Four separate biopsies were obtained at the same time but at different locations, during pre-treatment tumor resection, using a referenced three-dimensional data set generated by an ISG/Elekta Viewing Wand (ISG, Toronto, ON, Canada). Biopsy samples were chosen from the 3D data set corresponding to areas of tumor on T2-weighted magnetic resonance images (MRIs) and T1-weighted MRIs after gadolinium administration. Analysis of these biopsies revealed a heterogeneous pattern of LOH (Fig. 2 and Table 1, Case 1). Biopsy 1 showed LOH only at D1S199 and only 45% loss at this locus. In contrast, the other three biopsies showed essentially complete loss at both D1S214 and D1S199. Interestingly, all four biopsies were heterogeneous with regard to the degree of LOH seen at D1S2734, which varied among being maintained (Figure 2, Biopsy 1), marginal loss termed allelic imbalance (Biopsy 3), and LOH at 41% and 51% (Biopsies 2 and 4). That this variation in LOH at D1S2734 was not due to varying degrees of contamination by normal tissue in the tumor material is supported by histologic examination and by the observation that the other two markers showed essentially complete LOH in these same experiments. Similar data was obtained from analysis of multiple biopsies of a glioblastoma multiforme (Table 1, Case 2). In this case, D1S199 was not informative, but LOH at the other two markers was observed at none (Biopsy 1), one (Biopsies 3, 4, and 5), or both (Biopsy 2) remaining loci. Chromosome 1p LOH data from multiple biopsies of Case 1. Fluorescence intensity traces from capillary electrophoresis of LOH PCR samples derived from normal blood and multiple biopsies of Case 1 are shown. Arrows in the normal panel indicate the alleles of D1S199, D1S2734, and D1S214. Arrows in biopsy panels indicate alleles that have shown LOH (see also Table 1). Numbers above peaks indicate measured size of fragments as determined by the CEQ 2000XL Fragment Analysis (Beckman Coulter, Inc., Fullerton, CA) software on the basis of interpolation from size standards. Taken together, these data suggest that a tumor can show significant heterogeneity at the molecular level with regard to LOH on 1p. When the data is interpreted in light of the theory of clonal evolution, it is possible to suggest that the different patterns of LOH seen at different locations are evidence of progression. For example, it can be argued that in the oligodendroglioma (Table 1, Case 1) progression occurred in line with increasing levels of LOH at D1S2734, in the order Biopsy 1, 3, 2, and 4. Similarly, it is possible that in Case 2 the clone represented by Biopsy 1 gave rise to one represented by Biopsies 3, 4 and 5, and this evolved in to the clone at Biopsy 2. Accordingly, we do not interpret these data as evidence against clonal evolution. However, they do suggest that ancestral clones can persist to a significant degree in terms of the mass of the clinically evident tumor (Fig. 1b). Furthermore, when the marker used to assess heterogeneity has clinical implications, this heterogeneity may be of significance to treatment outcome. A clone that does not show LOH at 1p (for example, Case 2, Biopsy 1) may survive treatments that its clonal descendants, the more progressed 1p-negative clones (Biopsies 2 through 5), are eliminated by and so may be responsible for recurrence. Any molecular analysis that does not take into account this heterogeneity may not succeed in making clinically relevant correlations with outcome. In a second study, 30 breast cancers with multiple stages of disease in the same histologic section were studied for LOH at chromosome 18p11.3 (7). Following review by a pathologist, foci representing normal, ductal carcinoma in situ (DCIS), and invasive ductal carcinoma (IDC) were harvested from paraffin-embedded, sectioned, tumor material by microdissection. In some cases metastatic lesions also were available for analysis. DNA isolated from these tissues was analyzed for LOH using 18p11.3 markers D18S59 and D18S481 as well as a control marker, D18S452, at 18p11.2. The overall frequency of LOH found at 18p11.3 was 63%, which was significantly higher than levels previously reported using DNA isolated from bulk tumor. This finding underscores the importance of studying pure cell populations when investigating molecular characteristics of tumors. This is particularly important in all analyses addressing the clonality of tumor cell populations. More important for considering whether the theory of clonal evolution can explain the progression of these tumors was the finding that over 25% of them did not follow the classic progression paradigm (summarized in Fig. 3; for data see [7]). In seven cases, LOH at 18p11.3 was found in both DCIS and IDC, suggesting that the latter derived from the former. In another four cases, LOH was seen after the transition to IDC or to metastatic disease only, allowing that these tumors also progressed in a linear fashion. However, in two cases, LOH was observed in the DCIS but not in the IDC on the same slide, implying that the IDC component did not derive from the dominant clone observed in the DCIS. Similarly, in another two cases, microsatellite instability (MI) was observed in less aggressive lesions, and LOH was observed in more advanced stages of the same tumor. The fact that this LOH was not accompanied by any MI makes it unlikely that the more advanced forms resulted from the clones that showed MI. In summary, the evidence from LOH analysis did not support classic clonal evolution in just over 25% of cases studied. Future work on this series will attempt to isolate DNA from single cells in the DCIS of tumors that show complex clonal patterns (See top and middle cases on the right hand side of Figure 3.) and will attempt to see whether there are rare cells with LOH at 18p11.3 in the absence of MI. LOH data for 18p in breast cancer in the context of tumor progression and clonal evolution. Schematic representation of the data presented in (7), showing the stage at which LOH was detected in 18p in cases of breast cancer progression. IDC, invasive ductal carcinoma; DCIS, ductal carcinoma in situ. The data from the 1p analysis of oligodendrogliomas and 18p11 analysis of breast cancer suggest that a further refinement of the dominant clone interpretation of the theory of clonal evolution is warranted. In both studies, it appears that more than one clone can co-exist in some tumors. Furthermore, the breast cancer study provides direct evidence that the most aggressive clones, and therefore those of greatest clinical relevance, do not always arise from the previously dominant clone. It must be considered that, in a some tumors, multiple clones co-exist, from which the next step in the tumor's evolution can arise (Fig. 1c). Furthermore, the probability that a given clone is the one to give rise to the next progressed phase of the tumor is not necessarily simply a reflection of its prevalence. In other words, rarer clones, which may easily be overlooked in population-based analyses, could be as important for tumor progression, and so outcome and treatment choice, as the current dominant clone. This work was supported by the generosity of the Hermelin Brain Tumor Center Donors.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.