Ewing tumours and synovial sarcomas have critical features of karyotype evolution in common with epithelial tumours

Authors


Abstract

We have analysed the accumulated cytogenetic data on karyotypic evolution in Ewing tumours (ET) and synovial sarcomas (SS). Both tumour types frequently show balanced translocations, t(11;22) and t(X;18), respectively, that result in specific fusion genes. The analyses revealed +8, +12, +1q, and 16q− as important secondary changes to t(11;22) in ET and the imbalances showed a distinct temporal order. By principal component analysis, one major karyotypic pathway dominated by gains and one minor dominated by losses were identified. The kartyotypic evolution pattern in SS was less distinct. Both ET and SS showed a power law distribution of the number of acquired aberrations, which in both tumour types conformed to a distribution with an exponent equal to 1. Similar distributions are frequently found in epithelial tumours. ET and SS differ in this respect from other malignancies with balanced translocations resulting in fusion genes, which typically show a power law distribution of the number of acquired aberrations with exponents close to 2. This suggests that chromosome changes in ET and SS may develop through mechanisms more similar to those in epithelial tumours lacking recurrent balanced rearrangements than in haematological malignancies characterised by balanced translocations leading to fusion genes. © 2005 Wiley-Liss, Inc.

We have systematically analysed the structure of the karyotypic evolution in epithelial tumour types and identified several general features of their complex karyotypic patterns.1 First, the chromosomal imbalances are acquired in a preferred order. Second, two cytogenetic pathways are frequently seen in each tumour type, one characterised by gains and the other by losses, and third, the number of acquired chromosomal changes conforms to a power law distribution.1 To explain the observed distribution of the number of aberrations per tumour, we have proposed a model that includes a stepwise acquisition of aberrations that faithfully reproduces the power law distribution.2 In addition, a few tumour types, such as ovarian, head and neck, and lung carcinomas, develop through at least two phases of which only one (Phase I) is characterised by a power law distribution. We have taken these findings as an indication that the karyotype evolves in a highly ordered fashion during Phase I and that the transition from Phase I to Phase II involves a major change in the chromosome aberration dynamics.1 The change from Phase I to Phase II is associated with the onset of telomere crisis with the subsequent appearance of highly complex and disorganised karyotypes.3 In the present study, we have analysed Ewing tumours (ET) and synovial sarcomas (SS), both of which are characterised by specific balanced translocations resulting in fusion genes.4, 5, 6 Here we show that ET and SS have more features of the karyotype evolution in common with epithelial tumours lacking recurrent balanced rearrangements than with haematological malignancies characterised by balanced translocations leading to fusion genes.

Material and methods

Selection of data

All ET and SS with abnormal karyotypes were retrieved from the Mitelman Database of Chromosome Aberrations in Cancer.7 A total of 328 ET and 182 SS karyotypes were ascertained. The number of aberrations per tumour (NAPT) was estimated by scoring the number of entries in each karyotype. The NAPT value also includes balanced changes, e.g., t(11;22) and t(X;18). For comparison, NAPT were calculated in 442 acute myeloid leukaemias (AML) with t(8;21), 340 AML with 11q23 rearrangements, and in 1,746 Ph-positive chronic myeloid leukaemias (CML). All chromosomal breakpoints in ET and SS were registered and the frequency of breaks per band calculated.

To reduce the complexity of the data the karyotypes from ET and SS were used to construct imbalance maps, and the 21 chromosomal segments in ET and the 30 chromosomal segments in SS either lost or gained in more than 5% of the cases were identified (Table I). Each karyotype was then assessed for the presence, given the value 1, or absence, given the value 0, of the selected imbalances. The number of imbalances per tumour (NIPT) was calculated and the 290 cases of ET and 166 cases of SS with at least 1 of the selected imbalances were used for further analysis. This data was then used in the temporal analyses as well as in the principal component analyses.

Table I. Frequencies of Chromosomal Changes in ET and SS
Ewing tumorSynovial sarcoma
Chromosomal changeAbbreviationFrequency1 (%)Chromosomal changeAbbreviationFrequency1 (%)
  • 1

    Frequencies above 15% in bold.

+1q21–q44+1q20+2 12
+2 14+5q14–q35+5q5
+4 8+7 11
+5 9+8q11–q24+8q13
+6 6+9 8
+7 10+12 16
+8 36+15 7
+9 8+17q21–q25+17q6
+11q10–q25+11q6+18q11–q23+18q9
+12 22+21 9
+13q22–q34+13q6−1p36–p101p−15
+14 11−3p26–p143p−23
+15 8−4 10
+18 11−5 13
+20 12−6q14–q276q−11
+21 10−8p23–q128p−8
−7q32–q367q−6−9q32–q349q−5
−9p24–p139p−6−10 13
−11q24–q2511q−7−11 18
−16q11–q2416q−15−12p13–p1112p−8
−17 8−13 11
t(11;22) 72−14 13
   −15 12
   −16 9
   −17 13
   −18p11–q1118p−11
   −19 13
   −21p13–p1021p−7
   −22 14
   −Xp22–p11Xp−17
   t(X;18) 73

Temporal analysis

The time of occurrence (TO) was determined essentially as described previously.8 Briefly, all tumours with a given imbalance were selected and the distributions of NIPT plotted. Many of the distributions obtained in this way are skewed and the mean or median values are not appropriately characterizing the distributions. However, as modal values indicates when a given imbalance “typically appears”, the modes of these distributions were used as a value for TO. To obtain a better estimate of the TO, the selected distributions were resampled with replacement (bootstrapped) 1,000 times and the TO scored after each resampling.7 The mean of the bootstrapped TO values was then used as the TO for the given imbalance. The bootstrapped 25th and 75th percentiles were calculated and used as a measure for TO variability. For the bootstrap estimates, resampling software from Resampling Stats (Arlington, VA) was used.

Principal component analysis

To search for possible patterns of correlations between the imbalances, principal component analysis (PCA) was performed as previously described9 using the Statistica software package (Statsoft, Tulsa, OH). PCA is a standard multivariate method frequently used to search for underlying structures in data sets.10 In short, principal components are linear combinations of the original variables, orthogonal and ordered with respect to their variance so that the first principal component has the largest variance. Imbalance types were used as variables and the individual tumours as observations. This will group imbalances frequently seen in the same cases. By including the estimated TO values for each individual imbalance as an observation, the imbalances will be aligned along the first principal component according to their TO values. This is caused by the larger variability of TO, ranging from 1 to 14, than the presence or absence (1 or 0) of specific imbalances. The inclusion of TO in the PCA will therefore produce a representation in which the first principal component will correspond to a time axis.8

Statistical analyses

For statistical analysis and nonlinear estimations the Statistica software (Statsoft) was used. Probability density functions were fitted to the obtained distributions by use of the Levenberg-Marquardt algorithm.

Results

The NAPT values for ET and SS were calculated and plotted (Fig. 1a,c). The obtained distributions were similar and showed decreasing frequencies of cases with increasing NAPT values. As the distributions may be described by monotonously decreasing functions with heavy tails, the data were tested for meeting the requirements of a simple power law distribution, a generalised power law distribution (the Zipf-Mandelbrot distribution), an exponential, and a lognormal distribution, respectively. The observed data showed a best fit to the simple power law distribution, P(NAPT)= C × NAPT−α, with estimated α-values of 1.01 for ET and 1.03 for SS (Fig. 1b,d). This distribution is identical to the NAPT distributions of epithelial tumours1, namely, P(NAPT)= C × NAPT−1. We then compared the ET and SS NAPT distributions with those of other tumour types that show specific balanced chromosomal aberrations, namely, AML with t(8;21), AML with 11q23 rearrangements and Ph-positive CML. All neoplasms showed a power law NAPT distribution but with slopes steeper than what were seen in ET and SS (Fig. 2). The α-values for CML, AML with 11q23 rearrangements and AML with t(8;21) were 1.83, 1.93 and 2.2, respectively.

Figure 1.

The frequency distributions of number of aberrations per tumor (NAPT). (a) The ET NAPT frequency distribution. (b) A log-log transformation of the ET NAPT frequency distribution data with a fitted line. (c) The SS NAPT frequency distribution. (d) A log-log transformation of the SS NAPT frequency distribution data with a fitted line.

Figure 2.

The NAPT distribution of ET and SS compared to the NAPT distributions of renal cell carcinoma (RCC), AML with 11q23 rearrangements, AML with t(8;21) and CML. The log-log transformed fitted lines to the original NAPT distributions are given for each tumour type. The α values in p(NAPT) = C × NAPTα were 1.0, 1.0, 1.05, 1.8, 1.9 and 2.2 for ET, SS, RCC, CML, AML with 11q23 rearrangements, and AML with t(8;21), respectively.

We next investigated the distribution of chromosomal breakpoints across individual cytogenetic bands. The frequencies of breaks per band were rank ordered and plotted using the log transformed rank values (Fig. 3a). The ET data indicated the presence of 2 bands showing a high frequency of breaks, 22q12 (78%) and 11q24 (72%), 6 bands with frequencies higher than 3% and then the majority of bands with frequencies below 3% (Fig. 3a). The 6 bands with moderate frequencies were 22q11 (9%), 11q23 (6%), 1q21 (6%), 16q11 (5%), 1q12 (5%) and 16q13 (3%). The frequencies for the remaining bands conformed to an exponential distribution (Fig. 3b). A similar analysis of the breakpoint frequency distribution in SS (Fig. 3c) revealed the presence of 2 bands with a high frequency of breaks, 18q11 (80%) and Xp11 (80%). The frequencies for the remaining bands conformed to an exponential distribution and were in all cases below 7% (Fig. 3d).

Figure 3.

The frequency distributions of chromosomal breakpoints in ET and SS. (a) The ET breakpoint frequency distribution; insert shows the same distribution when the 2 first values are omitted. (b) The breakpoint frequency distribution in ET of breaks with ranks >8. The frequency distribution of these breaks conforms to an exponential distribution resulting in a straight line in the log-transformed data. (c) The SS breakpoint frequency distribution. (d) The breakpoint frequency distribution of breaks with ranks >2. The frequency distribution of these breaks conforms to an exponential distribution resulting in a straight line in the log-transformed data. The lines in (b) and (d) are not fitted lines.

The frequencies of the individual imbalances and the tumour specific translocations in ET and SS are shown in Table I. From Table I, it is evident that +8, +12, and +1q in ET are considerably more frequent (20–36%) than the other imbalances (15% or lower). The t(11;22)(q24;q12) was present in 73% of the cases. The temporal analysis of the imbalances showed that t(11;22) is the earliest change followed by 11q−, 16q− and +8, all with TO < 4, followed by the remaining imbalances (Fig. 4a). The temporal analysis of SS revealed t(X;18) to be an early change followed by Xp−, +8q, +7, and −22, all of which had TO values larger than 4.5 (Fig. 4b). The most frequent imbalances were Xp−, 1p−, 3p−, −11, and +12, all seen in 15% or more of the cases. Of these 1p−, 3p−, −11, and +12 were late changes with TO values larger than 7 in all cases.

Figure 4.

The temporal and PCA analyses of imbalances. Temporal analysis of (a) ET and b) SS. Circles in boxes indicate the mean of the bootstrapped TO; boxes show the boundaries of the 25th and 75th percentiles. Imbalances were first sorted according to the TO values and then imbalances with the same TO were organised with respect to the value of the 25th percentile. PCA analyses of the imbalances in (c) ET and (d) SS. As the first principal component (PC1) corresponds to a time axis, the karyotypic evolution is proceeding from the top to the bottom. For abbreviations see Table I.

The PCA of the imbalances in ET revealed one major and one minor cluster of imbalances (data not shown). By including the TO values as an observation in the PCA, the imbalances will be aligned to the first principal component according to their TO values and thus the first principal component will correspond to a time axis.8 The resulting 2-dimensional representation of the PCA suggests the presence of two possible pathways of imbalances (Fig. 4c), one starting with t(11;22), followed by +8, then by +12 or +14, and then by further gains, and one originating either with 16q− or 11q−, followed by +1q, then by −17, 9p− or 7q−, and then possibly by late gains in common with the previous pathway. A similar PCA of the imbalances present in SS revealed 2 clusters of imbalances, one dominated by gains and the other by losses, located at some distance from t(X;18) (Fig. 4d). The data thus indicate that after the acquisition of t(X;18), either chromosomal gains or losses are acquired without resulting in any distinct pathways.

Discussion

The analysis of the chromosomal breakpoints in ET revealed 6 bands in addition to 11q24 and 22q12 that were involved at distinctly higher frequencies then the bulk of breakpoints. Two of the bands, 11q23 and 22q11, can probably be explained by an imprecise cytogenetic classification of the 11q and 22q breaks in the t(11;22)(q24;q12). The remaining 4 bands, 1q21, 1q12, 16q11 and 16q13, could be associated with the der(16)t(1;16) frequently seen in ET.4, 5 The remaining breakpoints conformed to an exponential distribution, indicating the involvement of a stochastic process with no selection for specific breaks. The most frequent imbalances in ET were +8, +12, and +1q. The importance of these imbalances was also shown by their relatively low TO values; +8 and +1q belonged to the earliest events after t(11;22). Taken together, this identifies +8, +12, +1q, and 16q− as important changes in ET, as also pointed out by Sandberg and Bridge.5 The PCA revealed the presence of one major karyotypic pathway, including +8, followed by +12 or +14, and then by further gains, and one minor pathway characterised by either 16q−, 11q− or +1q, followed by −17, 9p− or 7q−, and then possibly by late imbalances also seen in the major pathway. The presence of two karyotypic pathways is a feature that ET has in common with several epithelial tumour types.1

The analysis of chromosomal breakpoints in SS did not reveal any significantly involved band in addition to Xq11 and 18p11, associated with the specific translocation t(X;18). The remaining bands showed breaks at substantially lower frequencies that conformed to an exponential distribution. SS differed from ET in that 4 of the 5 most frequent imbalances (1p−, 3p−, −11, and +12) were late changes, by the absence of distinct karyotypic pathways, and in particular by the absence of moderately late changes, i.e., imbalances with TO = 2–4. These findings could indicate that the secondary changes to t(X;18) are acquired in one single step, and hence that tumours with either only t(X;18) or with t(X;18) and many changes would be produced. However, the SS NAPT distribution conformed to a power law, indicating a stepwise acquisition of imbalances.2 Hence, the data rather indicate that after the acquisition of t(X;18), a number of alternative imbalances may be acquired as secondary changes, and that the subsequent imbalances may be acquired in almost any order, i.e., imbalances may function both as moderately late and as late imbalances. A consequence of this scenario is that t(X;18) cases with 1–3 additional imbalances would be very heterogeneous and hence no preferred moderately late imbalances would appear. The temporal analysis did however suggest that the temporal order was not completely random as this would have produced almost equal TO values for the different chromosomal changes, but this was not the case.

Despite the observed differences in the karyotypic evolution between ET and SS, the two tumour types share a fundamental feature, namely, the power law distribution of the number of acquired aberrations that in both tumour types conformed to the function P(NAPT) = C × (NAPT)−1. The two sarcoma entities share this feature with the majority of other solid tumour types,1 none of which are characterised by specific cytogenetic rearrangements. This finding is intriguing in as much as the balanced changes t(11;22) and t(X;18) were included when calculating NAPT for the ET and SS cases, and indicates that the presence of these translocations does not alter the general features of the chromosomal evolution. Interestingly, ET and SS differ in this respect from CML, AML with 11q23 rearrangements and AML with t(8;21), showing similar balanced translocations resulting in fusion genes, but typically having power law NAPT distributions with a steeper slope. This indicates a greater dependence on secondary changes in ET and SS than in AML, and CML.

In addition, the tumour-specific translocations in ET and SS behave similarly to early and frequent chromosomal imbalances present in tumour types that do not show any typical balanced translocation. For instance, 3p− is seen in 78% of clear cell renal cell carcinomas,11 which is comparable to the 73% of t(11;22) in ET and 72% of t(X;18) in SS. In addition, 3p− is seen in 64 % of the cases with single changes in RCC, which is comparable to 81% of t(11;22) among ET and 87% of t(X;18) among SS with sole anomalies. It is thus tempting to speculate that the t(11;22) and t(X;18), giving rise to the EWSR1/FLI1 and SS18/SSX1, SS18/SSX2, or SS18/SSX4 fusion genes,5 respectively, have similar impact on the mode of karyotypic evolution as changes caused by chromosomal imbalances in other solid tumour types.

We have previously shown that tumour types with complex karyotypes may develop through three phases: Phase I, II and III.1 Phase I tumours differ from Phase II/III tumours by having a lower number of imbalances, generally 1–10, by showing a temporally structured karyotypic evolution along certain pathways and by having a power law NAPT distribution with an exponent equal to 1. The finding that the NAPT distributions of both ET and SS conform to an identical power law distribution suggests that these tumour types develop in a fashion similar to Phase I epithelial tumours. Phase II and Phase III tumours, on the other hand, generally have more than 10 changes, do not show NAPT power law distributions and are distinguished by having highly disorganised karyotypes.1 Tumour types that show both Phase I and Phase II/III cases demonstrate subpeaks in their NAPT distributions. No such subpeaks were seen in the ET and the SS NAPT distributions, and hence both sarcoma types are homogenous with respect to the mode of karyotypic evolution. We have shown that the transition from Phase I to Phase II may be associated with the onset of telomere crisis.3 The absence of Phase II/III ET and SS tumours may suggest that these tumours have a stem cell origin, possibly associated with a maintained telomerase activity.12

Even though these findings strongly suggest that the structure of the karyotypic evolution in ET and SS is very similar to many epithelial tumours, they differ by demonstrating an exponential frequency distribution of the chromosomal breakpoints. Epithelial Phase I tumours show a power law distribution of the chromosomal breaks whereas Phase II/III tumours show an exponential distribution1. This may indicate that the breakpoint dynamics in epithelial Phase I tumours is different from the dynamics seen in ET and SS, and particularly that the breaks in ET and SS, except for those producing the specific fusion genes and possibly the +1q and 16q− in ET, are caused by stochastic processes that are not specific in the sense that they affect genes of importance for tumour development. Nevertheless, the critical features of the karyotype profiles of ET and SS is highly similar to the structure of karyotype profiles in many epithelial tumours, suggesting that secondary chromosome changes in ET and SS may develop through mechanisms more similar to those in epithelial tumours than those in haematological malignancies.

Ancillary