logo SBA


Digital archive of theses discussed at the University of Pisa


Thesis etd-02082013-230158

Thesis type
Tesi di dottorato di ricerca
Thesis title
Flow cytometry detection of neoplastic plasma cells: employing a new statistical model in diagnosing and monitoring Multiple Myeloma
Academic discipline
Course of study
tutor Prof. Petrini, Mario
  • flow cytometry
  • multiple myeloma
  • plasmacells
Graduation session start date
Multiple myeloma (MM) is a clonal B-cell disorder in which malignant plasma cells (PCs) accumulate in the bone marrow (BM), producing lytic lesions, excessive amounts of monoclonal protein in the serum or urine and evidence of end-organ damage (hypercalcemia, renal insufficiency, anemia or bone lesions). A conventional diagnosis in MM is based on a variety of laboratory results: morphology, analysis of M component, haematological features, biochemical parameters, immunophenotyping, cytogenetics, DNA ploidy and labelling index-proliferative activity of PC . Immunophenotypic studies on MM have now been performed for more than 15 years and flow cytometry (FC) represents an attractive approach, not only for research purposes but also in guiding clinical practice. In this sense, FC has many advantages: a) to distinguish among normal, reactive and malignant PC, b) to evaluate the risk of progression from monoclonal gammopathy of unknown significance (MGUS) to MM,, c) to detect prognostic markers,, d) to evaluate minimal residual disease (MRD) and e) to identify new targets for myeloma therapy. It may be difficult to define the clonality of a very small phenotypically abnormal plasma cell population in MRD, both by histology and FC. At this time histology remains the gold standard exam but we here define a new diagnostic model that is objective and reproducible at describing the correlation between phenotype and histology. Based upon our experience we found 8-color FC analysis to be a superior technique for identifying pathologic PC, especially in minimal residual disease (MRD). In a previous study we found that CD19 showed good correlation with the presence of disease using a cut-off of 61% to distinguish between normal and neoplastic PCs. However, this study evaluated the role of each antigen separately in detecting the presence of disease (this study was performed in Massachusetts General Hospital (MGH), Boston. We studied 15 control specimens and 55 patients). In another study (this study was performed in Pisa, at Hematology section, Santa Chiara Hospital, together with the collaboration of MGH. 15 control samples and 177 patients were studied. Patients were from both Santa Chiara Hospital and MGH) we demonstrated that the contribution of different antigens assessed simultaneously improved the correlation with histological results. We described a statistical model where, among all antigens tested, assessing CD19 and CD27 together resulted in the best concordance with histology. Its practical application is simple, rapid and does not require specialized technicians. We propose to use this formula as routine diagnostic tool. It could be used by a simple excel sheet or by a database, where, putting CD19 and CD27 expression values for each patient studied, a value of probability of disease can be obtained. A model plot could also be used for a quick test. A difference value of 0.2 could be used as cutoff of concordance between histology and model. Anyway, although if statistically acceptable, this model was found by analyzing samples collected in just two different laboratories and so, to improve the forecasting efficiency, our next goal will be to perform a multicenter study. Then we would like to evaluate the role that this flow cytometric model could have in evaluating response to therapy, in addition to the actual standardized criteria of evaluation. Thus, a multicenter national and international study, involving 21 Italian laboratories and laboratory of Patholgy of MGH, in Boston, is in progress. Until now, we have collected and analyzed MM samples from Hospital of Lucca. In the last year we have collected 64 MM samples from Lucca Hospital. We have tested the actual statistical model on these cases and we found that it works in 95% of cases. We then recalculated the statistical model, including samples from Lucca to those one from Pisa and Boston, and we found that CD19 (p-value=0.0000) and CD27 (p-value=0.0008) are still the two antigens best correlated with histological results, with an an R2 adjusted of 73,6%, so similar to that one previousely found. On all samples studied (120 from Pisa and Boston and 64 from Lucca), the percentage of concordant cases was 91.3%, with 4.7% discordant cases and 4% uncertain cases. Discordant cases were CD19 and CD27 positive myleoma cases and those cases where a monoclonal plasma cell population was detected together with a polyclonal plasma cell population by flow cytometry. Cases having uncertain results by the model have a difference between the value of the event observed by histology and the value of the event predicted by the model ranging between 0.20 and 0.46. These values are less than 0.50 and so could be considered acceptable, although, to make the model very restrictive, a value of difference between 0 and 0.20 was considered acceptable. This model makes predictions about the probability that the event “presence of disease” will occur for each patient, given the values of the variables CD19 and CD27.