There are several issues which must be worked out for MRD assessment to have broad utility in ES. First, it is unclear which site (blood or bone marrow) will ultimately provide the greatest clinical relevance. In patients with extensive tumor burden, assessment of either site is likely to yield the same result, although these patients will benefit the least from MRD testing because their disease is already clinically apparent. For patients diagnosed with initially localized disease, the impact of minimal bone marrow involvement on outcome has been inconsistent in smaller studies [5
]. However, results were more convincing in the largest trial to date [2
], which reported a decrease in 2-year disease-free survival from 80% versus 53% when bone marrow MRD testing was positive (P
= 0.043). It is possible that this may reflect that the impact on outcome is only apparent when a sufficiently large number of patients are tested. Another factor potentially leading to variable results is that bone marrow involvement in ES is more heterogeneous than that in leukemia, and it is common for morphology assessments of disease to differ between sides, and between the aspirates and core biopsies. This was evident in our institutional experience using FISH, in which one patient had aspirates from each side analyzed separately, with disparate results (3% versus 7%).
There is somewhat less data available regarding analysis of circulating tumor cells in ES. As with bone marrow a convincing effect on survival being related to circulating tumor cells at diagnosis is seen in larger [2
] but not some smaller studies [5
]. Collection of blood samples is far less cumbersome for patients than bone marrow, and is well suited for long-term monitoring either during or after completion of therapy. In fact, the latter approach may be particularly relevant, as several patients have been reported to have circulating tumor cells prior to clinically apparent relapse [5
]. In one of the larger studies, 10 of 11 patients with recurrence had tumor cells identified in blood or bone marrow by RT-PCR prior to overt relapse, with a median time lag of 4.5 months (range 1–24 months) [5
]. In our current trial, we are performing peripheral blood MRD evaluations any time patients undergo imaging assessments (at diagnosis, on therapy, or after therapy), while bone marrow testing is only performed when marrow samples would be routinely obtained for clinical purposes.
Quantification of RT-PCR results has not been generally reported, with the exception of Merino et al., who used real-time quantitative RT-PCR to estimate the effectiveness of a bone marrow purging method [15
]. It is possible that this approach would provide standardization of methodology and consistency in determining exactly what constitutes a positive test result. Similar standardization attempts would be helpful for flow cytometry, given the difficulties in interpreting results when there are only one or two events in the gated field.
Another question is whether cells identified by these methods are truly cancer cells, as each assay has the potential for false positives. Although RT-PCR detects pathognomonic EWS
changes not found in hematopoietic cells, contamination during RNA collection and testing may occur. For FISH, changes in the EWS
gene during decondensation of DNA can cause an occasional cell to appear as if there may be a true rearrangement, as discussed earlier and noted in . For flow cytometry, despite the use of a panel of markers to exclude hematopoietic cells, there is always the possibility of illegitimate transcription of these hematopoietic markers in tumor cells. In fact, in the most recent report by Ash et al. [22
], flow cytometry was reported to identify tumor cells in all 35 diagnostic bone marrow samples from patients with localized disease, and this incidence of 100% is in sharp contrast to all previous reports estimating the incidence of marrow micrometastases to be 20–30% in this patient population. Because the sensitivity of their assay is within the same range of that reported with RT-PCR, the question is raised whether all of these cells were indeed tumor cells. Efforts to reduce the potential for false positive results should continue.
Other methodologic issues include the specific protocols regarding how samples are collected and in what volume. Using a large volume (10
mL or perhaps more) may be ideal for collecting blood samples, particularly in patients who are on therapy and who may have treatment-related reductions in the number of circulating mononuclear cells. However, more may not necessarily be better with bone marrow collections, as demonstrated in a recent report by Helgestad et al. [26
]. They showed that the density of nucleated cells in the bone marrow of leukemia patients is markedly reduced with larger volume aspirates, due to potential dilution with peripheral blood during the collection. In fact, this dilution effect from larger aspirations resulted in several samples being interpreted as negative (below limits of sensitivity by flow cytometry), despite clearly containing >0.1% tumor cells in the first small volume sample withdrawn. Further, bilateral bone marrow aspirations are routinely performed for ES patients, due to the typically patchy tumor involvement. Most studies do not specify whether both sides are pooled together or analyzed separately. Attention to standardization of collection procedures will help improve interpretation of test results.
Finally, it remains unclear which assay has the greatest utility. Because of the success of flow cytometry for MRD assessment in leukemia, the readiness of commercially available antibodies, and the encouraging results noted so far in preliminary studies, it is likely that there will be further exploration of flow cytometry for MRD detection in ES. Results from ongoing trials which directly compare these methodologies will hopefully provide input on which assay to study in larger prospective clinical trials.