Reproducibility of the French-American-British classification of acute leukemia: the Southwest Oncology Group Experience

An Expandable Informatics Framework for Enhancing Central Cancer Registries with Digital Pathology Specimens, Computational Imaging Tools, and Advanced Mining Capabilities

BACKGROUND

Population-based state cancer registries are an authoritative source for cancer statistics in the United States. They routinely collect a variety of data, including patient demographics, primary tumor site, stage at diagnosis, first course of treatment, and survival, on every cancer case that is reported across all U.S. states and territories. The goal of our project is to enrich NCI's Surveillance, Epidemiology, and End Results (SEER) registry data with high-quality population-based biospecimen data in the form of digital pathology, machine-learning-based classifications, and quantitative histopathology imaging feature sets (referred to here as Pathomics features).

MATERIALS AND METHODS

As part of the project, the underlying informatics infrastructure was designed, tested, and implemented through close collaboration with several participating SEER registries to ensure consistency with registry processes, computational scalability, and ability to support creation of population cohorts that span multiple sites. Utilizing computational imaging algorithms and methods to both generate indices and search for matches makes it possible to reduce inter- and intra-observer inconsistencies and to improve the objectivity with which large image repositories are interrogated.

RESULTS

Our team has created and continues to expand a well-curated repository of high-quality digitized pathology images corresponding to subjects whose data are routinely collected by the collaborating registries. Our team has systematically deployed and tested key, visual analytic methods to facilitate automated creation of population cohorts for epidemiological studies and tools to support visualization of feature clusters and evaluation of whole-slide images. As part of these efforts, we are developing and optimizing advanced search and matching algorithms to facilitate automated, content-based retrieval of digitized specimens based on their underlying image features and staining characteristics.

CONCLUSION

To meet the challenges of this project, we established the analytic pipelines, methods, and workflows to support the expansion and management of a growing repository of high-quality digitized pathology and information-rich, population cohorts containing objective imaging and clinical attributes to facilitate studies that seek to discriminate among different subtypes of disease, stratify patient populations, and perform comparisons of tumor characteristics within and across patient cohorts. We have also successfully developed a suite of tools based on a deep-learning method to perform quantitative characterizations of tumor regions, assess infiltrating lymphocyte distributions, and generate objective nuclear feature measurements. As part of these efforts, our team has implemented reliable methods that enable investigators to systematically search through large repositories to automatically retrieve digitized pathology specimens and correlated clinical data based on their computational signatures.

The predictive value of morphological findings in early diagnosis of acute myeloid leukemia with recurrent cytogenetic abnormalities

This study explores cytomorphologic features and their predictive role for early identification of acute myeloid leukemia (AML) with morphological distinctive recurrent cytogenetic abnormalities (RCA): t(15;17), t(8;21) and inv(16)/t(16;16). We retrospectively evaluated 396 de novo AML cases, diagnosed and treated at single institution, between 2013-2017. Specific cytomorphologic features suggesting distinctive AML-RCA were revealed at diagnosis in 62 (15.65%) patients, including AML with t(15;17) in 41 (66.13%), t(8;21) in 13 (20.97%) and inv(16)/t(16;16) in 8 (12.90%). Final diagnoses of AML-RCA according to WHO integrated diagnostic criteria were established in 66 (16.66%) cases, including AML with t(15;17) 40 (60.60%), t(8;21) 17 (25.76%), and inv(16)/t(16;16) 9 (13.64%). Discordance between cytomorphological and other integrated criteria was detected as missed/wrong-call in 0/1 for t(15;17), 6/2 for t(8;21) and 2/1 for inv(16)/t(16;16). The cytomorphological accuracy was 97.56% (40/41) for t(15;17), 57.89% (11/19) for t(8;21) and 70% (7/10) for inv (16)/t(16;16). Positive/negative predictive values of cytomorphological evaluation were: 97.56%/100% for t(15;17); 84.62%/88.68% for t(8;21); 87.50%/96.65% for inv(16)/t(16;16). Sensitivity/specificity were: 100%/96.15% for t(15;17); 64.10%/95.92% for t(8;21); 77.78%/98.25% for inv(16)/t(16;16). We confirmed that morphology is still a highly relevant evaluation method in diagnosing several common AML-RCAs before completing cytogenetic and molecular studies, enabling early detection, particularly of AML with t(15;17).

Acute Myeloid Leukemia Genetics: Risk Stratification and Implications for Therapy

Acute myeloid leukemia is a category of diseases with a common aggressive clinical presentation but with a prognosis and management that is dependent upon the underlying genetic characteristics of the neoplasm. The purpose of this brief review is to update the practicing pathologist on the current standard of care in the genetic evaluation of acute myeloid leukemia and to highlight future directions in the classification, genetic assessment, and management of these devastating diseases.

Image mining for investigative pathology using optimized feature extraction and data fusion

In many subspecialties of pathology, the intrinsic complexity of rendering accurate diagnostic decisions is compounded by a lack of definitive criteria for detecting and characterizing diseases and their corresponding histological features. In some cases, there exists a striking disparity between the diagnoses rendered by recognized authorities and those provided by non-experts. We previously reported the development of an Image Guided Decision Support (IGDS) system, which was shown to reliably discriminate among malignant lymphomas and leukemia that are sometimes confused with one another during routine microscopic evaluation. As an extension of those efforts, we report here a web-based intelligent archiving subsystem that can automatically detect, image, and index new cells into distributed ground-truth databases. Systematic experiments showed that through the use of robust texture descriptors and density estimation based fusion the reliability and performance of the governing classifications of the system were improved significantly while simultaneously reducing the dimensionality of the feature space.

Relevance of pathologic classifications and diagnosis of acute myeloid leukemia to clinical trials and clinical practice

Many new insights into the diagnosis, pathogenesis, clinical manifestation, treatment and prognosis of patients with AML reflect the heterogeneity of the disease. The initial descriptions of the various subtypes of AML, established by the FAB classification, were based on morphology and cytochemical stains. Although morphology remains the foundation for the diagnosis, additional diagnostic studies including immunophenotyping, cytogenetic evaluation, and molecular genetic studies have become critical, and in some specific cases, mandatory, complementary tools. Several specific subtypes of AML are now treated with directed or targeted therapy. Acute promyelocytic leukemia is currently the only example of a subtype of AML to which specific therapy targeted to a molecular genetic abnormality is available and this subtype now is highly curable. Future studies will address newly identified prognostic factors and gene mutations such as FLT3, Wilm's tumor (WTI), and CEBPA which will enable the further pathologic classification of patients with AML. Finally, microarray analysis will likely identify genes critically involved in the pathogenesis of specific pathologic subtypes.

Low expression of the myeloid differentiation antigen CD65s, a feature of poorly differentiated AML in older adults: study of 711 patients enrolled in ECOG trials

CD65s appears when the progenitor antigen CD34 disappears, suggesting that this sialylated carbohydrate antigen marks a turning point in normal myeloid differentiation. We characterized acute myeloid leukemia (AML) with low CD65s expression (CD65s(low) AML) in 711 patients entered on seven Eastern Cooperative Oncology Group AML treatment trials (1986-1999). Of those, 198 (28%) qualified as having CD65s(low) AML. Morphologically, CD65s(low) AML was more common in FAB subgroups with minimal differentiation, M0/M1 (P=<0.0001). Early precursor antigens CD34, CD117 and terminal transferase were more frequent in CD65s(low) than CD65s(high) AML (P=<0.0001). Myeloperoxidase was present in fewer CD65s(low) myeloblasts, and the more mature myeloid antigens, CD15 and CD11b, were rarely detected (P=<0.0001). Yet, the two diagnoses did not differ in the distribution of cytogenetic prognostic groups or the occurrence of the multidrug-resistance mediator, P-glycoprotein. CD65s(low) AML patients were significantly older than CD65s(high) cases (P<0.0001). Furthermore, the incidence of CD65s(low) cases increased with age, from 20% in patients under the age of 50 years to 67% in patients older than 80 years (P<0.0001). Overall, complete remission (CR) rate and overall survival were comparable in CD65s(low) and CD65s(high) AML. However, among patients >55 years of age, CD65s(low) AML had a decreased CR rate of 33 vs 44% in CD65s(high) AML (P=0.055). Thus, CD65s(low) AML represents immunophenotypically undifferentiated disease and occurs predominantly in older adults. Although not statistically significant, the observed association between low CD65s expression and decreased CR rate only in patients over the age of 55 is intriguing.

Computer-assisted discrimination among malignant lymphomas and leukemia using immunophenotyping, intelligent image repositories, and telemicroscopy

The process of discriminating among pathologies involving peripheral blood, bone marrow, and lymph node has traditionally begun with subjective morphological assessment of cellular materials viewed using light microscopy. The subtle visible differences exhibited by some malignant lymphomas and leukemia, however, give rise to a significant number of false negatives during microscopic evaluation by medical technologists. We have developed a distributed, clinical decision support prototype for distinguishing among hematologic malignancies. The system consists of two major components, a distributed telemicroscopy system and an intelligent image repository. The hybrid system enables individuals located at disparate clinical and research sites to engage in interactive consultation and to obtain computer-assisted decision support. Software, written in JAVA, allows primary users to control the specimen stage, objective lens, light levels, and focus of a robotic microscope remotely while a digital representation of the specimen is continuously broadcast to all session participants. Primary user status can be passed as a token. The system features shared graphical pointers, text messaging capability, and automated database management. Search engines for the database allow one to automatically identify and retrieve images, diagnoses, and correlated clinical data of cases from a "gold standard" database which exhibit spectral and spatial profiles which are most similar to a given query image. The system suggests the most likely diagnosis based on majority logic of the retrieved cases. The system was used to discriminate among three lymphoproliferative disorders and healthy cells. The system provided the correct classification in more than 83% of the cases studied. System performance was evaluated using rigorous statistical assessment and by comparison with human observers.

Acute myelogenous leukemia: current treatment and future directions

Acute myelogenous leukemia (AML), although an uncommon disorder, is a useful prototype for the treatment of malignancies in general. Significant advances have been made in both the understanding and treatment of this disease. In particular, clinically relevant molecular mechanisms of disease in AML are being defined that hold future therapeutic promise. We review the classification and biology of AML and the current treatment controversies in the use of chemotherapy and bone marrow transplantation, and suggest directions for future research.

Quantitative enzyme determination in acute myeloid leukemia

Freshly obtained leukemic cells from 64 patients with acute myeloid leukemia were investigated. Intracellular protein and enzyme content (myeloperoxidase (MPO), acid esterase with alpha naphthyl acetate (ANAE) and alpha naphthyl butyrate (ANBE) and lactate dehydrogenase) were measured together with the cell diameter and compared with FAB-classification, cytochemical staining and clinical features. Mean values of MPO and acid esterase discriminate between the pure myeloid leukemias, although overlap between various subgroups is considerable. No correlation between cytochemical and quantitative enzyme determination was found. The logistic regression analysis revealed a dependency of the occurrence of complete remission on protein and ANBE (p less than 0.01) content. We concluded that quantitative enzyme determination is an easily measurable parameter of maturation and may have prognostic significance.

Acute nonlymphocytic leukemias of childhood. Inter-observer variability and problems in the use of the FAB classification

The French-American-British (FAB) classification system and some recent modifications were applied to 486 children with a diagnosis of acute nonlymphocytic leukemia (ANLL) to determine the distribution of the subtypes in children, to document the extent of inter-observer variation in assigning subtypes, and to examine the reasons for the differences. The distribution of FAB subtypes of childhood ANLL was similar to that reported for adults. In the initial year of the study, the inter-observer concurrence between the institutional diagnosis and the reviewing pathologists was 50%, but in the more recent years, concurrence between institutions and the review pathologist has approached 80%, averaging around 73% for the entire study. Many problems remain to be solved with this classification system, including the imprecision in wording, the subjectiveness of the interpretation, errors due to the random distribution of cells, and the current lack of evidence that certain FAB subtypes, such as M1 and M2, differ significantly in terms of biological behavior and prognosis.

Routine immunophenotyping of acute leukaemias

Recent progress in immunophenotyping includes the availability of monoclonal antibodies (MAbs), knowledge of specificity and reactivity patterns of these reagents, and the technical improvements and standardization of immunofluorescence and immunocytology staining procedures, including flow cytometry. These advances have contributed significantly to the establishment of immunophenotyping as an essential diagnostic tool in the differential diagnosis of types of acute leukaemia. Immunophenotyping allows for the objective and reproducible distinction of acute lymphoblastic leukaemia (ALL) from acute myeloblastic leukaemia (AML) and of T-lineage from B-lineage ALL. Immunologically defined ALL and AML subtypes have been found to convey prognostic significance. Using cell lineage-specific and differentiation stage-specific MAbs, cases of T- and B-lineage ALL and of AML can be further classified into a number of different subtypes. Routine immunophenotyping concentrates on the diagnostic enquiry into a few major, clinically relevant subtypes; only a limited number of crucial reagents are employed that are commercially available. The simplification and standardization of discriminatory immunomarker panels make immunophenotyping a reliable diagnostic instrument for the provision of critical data to make a differential diagnosis. An effort to identify the nature and origin of the blast cells precisely, immunological typing definitely plays an important part in the multiple-marker analysis of acute leukaemia (morphology, cytochemistry, karyotyping, genotyping) for applied diagnostic and fundamental research purposes.

Flow cytochemical patterns of white blood cells in human haematopoietic malignancies. I. Acute leukaemias

Peripheral blood samples from 118 patients with acute leukaemia (68 untreated; 50 treated) were measured with the Technicon H-6000 automated haematology analyser. This instrument provides, in addition to measurements of the classical haematology parameters (i.e. cell counts, haemoglobin concentration, etc.), a differential count on 10(4) WBC effected by means of flow cytochemistry (peroxidase content) and volume (light scatter) discrimination. Disregarding RBC and platelet counts and their volume distribution profiles, the most important diagnostic parameters for leukaemic disease were the WBC count, the WBC differential count, and the proportions of large unstained cells (LUC) and high peroxidase (HPX) cells obtained by the automated differential count as well as the mean value of the WBC peroxidase content distribution (MPA). Granulocytic leukaemias had lower MPA than normal and lymphocytic leukaemias had MPA values above normal. M1 leukaemias were also characterized by large proportions of LUC and low fractions of HPX, while M2 leukaemias showed low LUC with high HPX. M3 leukaemias had low LUC and very high HPX. M4 leukaemias had large LUC and 'monocytic' components and a modest fraction of HPX. M5 leukaemias had very large numbers of LUC, 'monocytes' and 'lymphocytes' and a normal HPX. For M1 leukaemia, the presence of less than 7% LUC following induction treatment was related to morphological changes of normal cells induced by chemotherapy while LUC above 10% usually indicated unsuccessful induction associated with the presence of residual blasts. If treatment was successful, M2 and M3 leukaemias characteristically decreased their HPX population. All M4 leukaemias studied by us failed to enter remission and continued to display high proportions of HPX and LUC. Similarly, most M5 leukaemias had a poor response to treatment and always showed a very high proportion of LUC. Untreated lymphocytic leukaemias demonstrated high LUC, normal HPX and a high proportion of 'lymphocytes'. Hairy cell leukaemias showed almost equal proportions of 'lymphocytes' and LUC. Successful chemotherapy of all lymphoid leukaemia entities was associated with rapid decreases in LUC, slower decrements of 'lymphocytes' and moderate and transient increments in HPX. Thus, flow cytochemistry can assist not only in the segregation of acute leukaemias along with FAB classification with nonmorphologic criteria, but also in the follow up of patients with these diseases.

Occurrence of particular isoenzymes in fresh and cultured leukemia-lymphoma cells. III. Esterase isoenzyme in monocytes

The expression of a particular alpha-naphthyl acetate esterase isoenzyme which is specific for monocytes was examined in a panel of cultured leukemia-lymphoma cell lines (n = 88), freshly obtained leukemia-lymphoma cells (n = 527), and in fresh (n = 10) and cultured (n = 22) leukemia cells treated with the phorbol ester 12-O-tetradecanoylphorbol 13-acetate (TPA). The sodium fluoride-sensitive isoenzyme was separated by isoelectric focusing on horizontal thin-layer polyacrylamide gels. The esterase isoenzyme was not detected in untreated or TPA-treated lymphoid, erythroid, or Hodgkin's disease-derived cell lines, but was seen in leukemia cell lines of monocytic origin. TPA induced the new expression of this marker isoenzyme in two leukemia cell lines of promyelocytic and erythroid origin that are known to differentiate along the monocytic-macrophage cell lineage; TPA stimulation increased the staining intensity of the band in monocytoid cell lines. This esterase isoenzyme was found in 92% of the cases classified morphologically as acute myelomonocytic or monocytic leukemia, but only in 3% of the non-monocytic acute myeloid leukemias. All lymphoid or erythroid leukemias or lymphomas were negative. Treatment with TPA of AML and CML cells, which commonly differentiate to monocyte/macrophage-like cells, showed de novo the monocyte-specific isoenzyme. It is concluded that this isoenzyme is a characteristic marker for monocytic leukemia cells and will be a useful tool for the discriminatory identification of the monocytic element in normal and leukemic cells.

Two-dimensional gel electrophoresis of cellular proteins reveals myeloid origin of blasts in two children with otherwise undifferentiated leukemia

In previous studies, two-dimensional polyacrylamide gel electrophoresis allowed the detection of 11 polypeptide markers that distinguished between subtypes of acute lymphoblastic leukemia and between acute lymphoblastic and acute myelocytic leukemia. This report describes the occurrence of polypeptide markers in two-dimensional gels that indicate a myeloid origin of blasts obtained from two children who presented with acute leukemia, the cells of origin of which could not be determined, at the time of diagnosis, by morphologic, cytochemical, or immune marker analysis. The authors conclude that two-dimensional electrophoresis provides a new tool for the delineation of the cell of origin in acute leukemia.

Monocyte-associated acid phosphatase isoenzyme profiles as determined in acute myeloid leukaemia cells

The acid phosphatase (acP) isoenzymes from the blast cells of 102 cases of acute myeloid leukaemia were separated by isoelectric focusing on horizontal polyacrylamide gels. The cases were classified on the basis of the FAB cooperative group criteria. Several single bands were combined into groups (I-IV). An increase in the number of acP isoenzymes was noted which paralleled the assumed maturation along the granulocytic cell lineage from FAB M1 to FAB M3 and along the monocytic cell lineage from FAB M4 to FAB M5. One isoenzyme which was resistant to tartrate inhibition was found in 40% of the monocytic variants FAB M4 and M5, but not in the nonmonocytic cases FAB M1-M3 and M6. This particular isoenzyme, which has been described as being characteristic for hairy cell leukaemia, also appears to be a marker of the monocyte/macrophage system and the respective neoplastic counterparts. The FAB M4 and M5 patients expressed a characteristic profile of group I isoenzymes which allows for the discrimination between monocytic and nonmonocytic cells.

Expression of a monocyte-specific esterase isoenzyme in cases of acute myeloid leukemias

The carboxylic esterase (E.C. 3.1.1.1) isoenzymes from cases of acute myeloid leukemias were separated by analytical isoelectric focusing on horizontal thin-layer gels. One isoenzyme consisting of one or two components (bands) could be completely and selectively inhibited by addition of 40 mM sodium fluoride (NaF) to the staining bath. The 105 cases were classified into the groups M1-M6 according to the FAB proposals. The NaF-sensitive isoenzyme was not detected in cases of FAB groups M1/2 (acute myeloblastic leukemia without or with maturation), group M3 (acute promyelocytic leukemia) or group M6 (erythroleukemia). Thirty-one out of 33 cases in the FAB group M4 (acute myelomonocytic leukemia) and 9/9 cases in FAB group M5 (acute monocytic leukemia) expressed the NaF-sensitive isoenzyme. The NaF-sensitive isoenzyme was found at different staining intensities; all M5 cases showed the isoenzyme at strong or very strong intensity, whereas most of the M4 cases displayed the isoenzyme at weak, medium or strong staining intensity. The data presented are further evidence that the presence of the NaF-sensitive esterase isoenzyme indicates monocytic involvement or differentiation in cases of myeloid leukemias. The easy and fast to perform method of isoelectric focusing can be used to distinguish the monocytic variants among the acute myeloid leukemias and can supplement the morphological analysis of these cases.

Institutional performance in application of the FAB classification of acute leukemia. The Southwest Oncology Group experience

Institutional performance in application of the French-American-British (FAB) classification of acute leukemia in The Southwest Oncology Group is presented, demonstrating a disparity between institutional and expert performance. A significant improvement is shown with an educational effort coupled with experience in use of the classification, and the importance of cytochemistry in the use of the classification is illustrated. A simplification of the classification, merging M1, M2, and M4 as M7, is proposed. This simplification improves institutional performance in use of the classification, while preserving distinctions that appear clinically important.