The outcome of patients with APL has improved over the last 2 decades as a consequence of a precise molecular diagnosis, better supportive care as well as the use of trans-retinoic acid (ATRA) and arsenic trioxide in chemotherapy protocols [30, 34]. Nevertheless, 20% of the patients still die during the first month of treatment, mainly due to bleeding. So, it would be interesting to detect the clinical and laboratory features that could predict a shorter overall as well as disease-free survival. Peripheral leukocyte counts above 10x109/l and a microgranular morphology have been recognized as such variables [2–5]. Additional chromosomal abnormalities, however, do not seem to influence the outcome of the patients [6, 7]. Therefore, it would be interesting to study the relation of these features with molecular characetristics of this disease. Nuclear texture features of neoplastic cells have shown to be morphological correlates of chromatin remodelling with genetic and epigenetic alterations in acute leukemias [12–17], and have shown to be related to patients’ survival. Recently, computerized image analysis has been used to examine chromatin remodeling in several hematological neoplasms and texture variables of the nuclear architecture have shown to be independent prognostic factors in acute leukemias [16–18, 23, 26, 27, 29].
In the present study, in a rather small cohort of patients, diagnosed and treated according to the international standard of care, we could observe a mortality rate of 21% during the first month of treatment which was similar to that observed in other countries [4, 30, 34]. We could also confirm the prognostic value of peripheral leukocyte counts and microgranular morphology.
More recently, several investigations have focused on the impact of mutations of the FLT3 gene in APL [2, 3, 5, 7, 9–11, 35] that has been a well recognized adverse prognostic factor in AML with a normal karyotype. In APL, the finding of FLT3-ITD has been associated with increased peripheral leukocyte counts, higher early mortality and more frequent recurrences. The presence of FLT3-ITD in APL is also associated with a different expression of 147 genes involved in cytoskeleton organization, cell proliferation and migration, adhesion, as well as the coagulation and inflamation pathways . Thus, it is linked to a more aggressive clinical behaviour with higher peripheral leukocytosis and a more pronounced coagulopathy, provoking a higher rate of early death, mainly due to bleeding. The higher leukocytosis is also indicative of a more pronounced proliferative activity. Our study also corroborated these findings. FLT3-ITD was a significant adverse prognostic factor for survival in the univariate Cox-regression. Furthermore, other variables, such as peripheral leukocyte count, microgranular morphology and texture features of nuclear chromatin were significant prognostic factors only when examined isolately, but lost their importance when the regressions were stratified for the presence of FLT3-ITD. From these findings we may hypothesize that the presence of the FLT3-ITD mutation might be more important for a dismal outcome than leukocytosis alone or a microgranular morphology. Additional studies with larger number of patients, however, are necessary to clarify this question.
Neither the FLT3-TKD mutation nor the methylation of CDNK2B, were unfavourable prognostic features in our investigation. Yet, a possible role of CDNK2B for the outcome of patients with other types of acute leukemias has been postulated . Bit this could not be confirmed for APL in the present study, perhaps due to the reduced test power caused by the small cohort of patients.
Recently, the importance of epigenetic changes for normal hemopoietic maturation as well as in the pathogenesis of AML has been emphasized [14, 17–19]. Hypermethylation of CpG islands within the promoter regions together with deacetylation of certain histones is an important mechanism of gene silencing in hematologic neoplasias. Several studies have shown that APL is generally associated with a specific methylation pattern, but some genes, such as CDKN2B, may be methylated isolately in only a part of the patients [15, 19]. Patients with the FLT3-ITD mutation show a gene expression profile where various genes relevant for cytoskeleton organization, cell adhesion and migration, proliferation and coagulation pathways are expressed .
MGG staining of cytologic smears permits to evaluate the topographic localization of methylated regions in the nucleus, since deeply Giemsa-stained compact heterochromatin domains are co-localized with methyl-rich CPG islands . Computer-assisted image analysis is able to document discrete changes in the chromatin structure, which may not be visible for the trained observer [38–43]. Discrete morphologic alterations of the nucleus and its substructures accompany functional and molecular changes of the cell. This is also true for benign and malignant hemopoesis [44–48].
High order chromatin architecture alterations may parallel chromosomal alterations in cancer [44, 45]. Moreover, disrcrete morphologic changes of the chromatin texture may be equivalent to alterations of the methylation pattern and therefore of the gene expression. Probably for these reasons computerized texture analysis of chromatin has shown to be helpful for diagnosis as well as for prognosis of several neoplasias including AML [16–18, 23–28, 38–43]. In particular, a fractal model of nuclear chromatin has been associated with cellular activity, the organization of nuclear chromatin and the surrounding nucleoplasmic space, especially the distribution of heterocromatin, and with prognosis [16, 25–29, 46].
In our study several variables of geometric morphometry, the gray-level co-ocurrence matrix and also R2 showed different values in leukemic cells of cases with and without the FLT3-ITD mutation. In patients presenting FLT3-ITD, cells had a smoother and more homogeneously distributed chromatin which would more likely correspond to profound alterations of the epigenome. The presence of FLT3-ITD has shown to be associated with a higher proliferative activity and a more undifferentiated phenotype . This is in keeping with the different chromatin texture features found in our patients as well as the higher PB leukocytosis.
Several clinical, phenotypical and nuclear chromatin texture features were associated to the survival of our patients. However, when stratified for the presence of FLT3-ITD, all of them lost their prognostic value. In that way, our explorative study indicated that this mutation is a very important prognostic factor in APL. This is in keeping with the concept of a previous study that APL with FLT3-ITD mutation represents a distinct subtype of APL with a worse prognosis .
In conclusion, our work underlines the necessity of the detection of the FLT3-ITD mutation in APL, that constitutes a separate entity of a worse prognosis and where new treatment strategies are necessary in order to decrease early mortality in an otherwise highly curable form of acute leukemia. Texture analysis demonstrated that the FLTD-ITD alterations were accompanied not only by different cytoplasmic features, but also by a change in the chromatin structure in routine cytologic preparations. Yet we were not able to detect chromatin changes by nuclear texture analysis of patients with the FLTD-TKD mutation or methylation of specific cell genes.