Acute Myeloid Leukemia: A general introduction

Salvatore Raieli
Oct 9, 2020 · 10 min read

In this article I will describe briefly what AML is, the current classification, current therapies available and open questions. In the next articles, I will discuss about diagnosis, risk classification with a particular eye about machine learning models and approach to them. Some of the presented approaches have been developed for other leukemia subtype, but the principles are the same and can be applied to AML. In the next articles, it will be a focus also on limitations and perspectives in leukemia.

Acute myeloid leukemia (AML) is a serious disease (where more of 20 % of the blood cells are cancerous cells or myeloblasts), and if left untreated is becoming rapidly fatal. Diagnosis is often obtained by noticing abnormalities in the blood count during a blood test. The diagnosis is completed by microscopical analysis of blood smear and often bone marrow aspiration (microscopical or flow cytometry analysis). AML arises from disorder of hematopoietic stem and progenitor cells (HPSCs), this is generally caused by genetic abnormalities (inherited or acquired). Despite the fact that it can occur at any age; an AML diagnosis at elderly age is considered more likely. It is considered one of the deadliest cancers (5-year overall survival inferior to 30%) and with high relapse rate and high rate of resistance to the therapy. AML is leading to primary infiltration of the hematopoietic organs: bone marrow, spleen and lymph nodes. This is resulting life-threatening complications such as immunodeficiency, thrombocytopenia and anemia (1). Despite the emerge of new technologies and insight on AML biology the standard therapy has not changed much in the last decades.

Image for post
Image for post
Figure 1. HPSCs stage where AML can develop. Source: (2).
Image for post
Image for post
Figure 2: example of blood smear. Figure source: (3).

Traditionally, AML has been classified according to morphology and the immunophenotype of the myeloblasts. Since the advent of omics technologies, it has been questioned the idea that AML is a homogeneous disease: current view considers AML as a heterogeneous group of neoplastic disorders. The new classification is taking in account the genetic anomalies, for instance chromosome translocation, fusion genes and other abnormalities. These genetic abnormalities have a great impact on the prognosis and has to be take in to account in the choice of the treatment protocol. World Health Organization (WHO) has nowadays classified AML in 6 different categories or subtypes (4):

1. AML with recurrent genetic abnormalities. It accounts for around 20–30 % of patients diagnosed with AML and presents a particular profile of chromosomal translocation or inversions.

2. AML with myelodysplasia related changes. Generally patient with myelodysplastic syndrome which evolve in AML

3. Therapy-related myeloid neoplasm. AML disease which evolves as an effect of chemotherapy or radiotherapy.

4. Acute myeloid leukemia, not otherwise specified. AML cases which are not classified in the other categories

5. Myeloid sarcoma. A cancer mass of myeloblasts which is localize in a different site from the bone marrow

6. Myeloid proliferations associated with Down syndrome. Individuals who have Down syndrome are known to have an increased risk of AML.

There are also other classification systems which are based on prognosis and risk stratification but they are also taking in account genomic alterations.

The AML therapy is composed of two phases: induction (with the scope of complete remission or reducing the myeloblast cell to an undetectable level) and consolidation (to eliminate the undetectable residual of leukemia to prevent an early relapse). Conceptually, consolidation therapy is aimed to eradicate the residual myeloblast and cancer stem cells. Young AML patients are treated with intensive chemotherapy or according the prognosis with bone marrow transplantation (normally transplantation is used in the case of induction therapy is ineffective). However, more than 50 % of AML newly diagnosed patients are older than 65 years, which are not eligible for an aggressive chemotherapy and presents a poor prognosis (5-year survival probability less than 10 %). Indeed, there is the need to develop new therapies for a large part of the AML of patients (5). Latest studies (conducting next generation sequencing) showed that a short list of gene mutations occurs in the majority of the AML patients (for instance: FLT3, NPM1, DNMT3, IDH1 and IDH2) (6,7). Moreover, other studies have also noticed mutation in WT1 and TP53 which are tumor suppressor genes. These mutations can be also predictive for the outcome: for example, patients with NPM1 mutation are more responsible for chemotherapy while patients with NPM1 and FLT3 mutation have a less favorable prognosis with increased risk of relapse (8).

Image for post
Image for post
Figure 3. the circular plot is showing the relative frequency and pairwise co-occurrence of genetic alterations in a large cohort. Figure source: (9).

New therapies have been developed based on these mutations which have been approved by the FDA in the last years (for example, drug against FLT3 or IDH mutated AML). Since FLT3 mutation leads to more aggressive phenotype and lower survival, different research groups have been interested in the research of effective drug against FLT3. AML patients have benefited from first and second generation of FLT3 inhibitors with improved survival. However, the patients which present these mutations and meet the criteria for these new treatments are about 50%, leaving the need of new treatments for the excluded patients. Moreover, since many of these patients are elderly, we need drugs that are less toxic (more tolerable) and could be combined with other drugs (conventional or not)

Among the new therapies in development there are drugs which are acting on specific signal pathways. As an example, there are drugs targeting apoptosis through BCL-2 inhibition. A BCL-2 is a gene that promotes cell survival regulating the apoptotic pathway and has shown to play an important role in myeloblast resistance to chemotherapy. For these reasons, there is in study a selective inhibitor of BCL-2. Another treatment in evaluation is against MDM2 which is another gene implicated in the escape from apoptosis. A part from apoptosis there are study evaluating drug directed to proliferation, differentiation and other pathways. As other examples, there are drug in development against the hedgehog pathway, CDK9 and so on (10).

Another approach is based on the epigenetic regulation of DNA. Gene expression in the nucleus is regulated by chemical modification of the DNA such as methylation or acetylation of the histone (proteins bounded to the DNA). The latest researches have shown that these processes are unbalanced in the myeloblasts. DNMT3, IDH1 and IDH2 mutation are leading to methylation unbalance in AML. FDA has approved a Azacitidine or Decitabine that has the first class of epigenetic drug, these drugs showed to inhibit the methylation of DNA. However other drug targeting epigenetic processes are in evaluation as BET inhibitors, LSD1 inhibitors and DOT1L inhibitors (10).

Image for post
Image for post
Figure 4: molecular alteration and signaling pathway. Figure source: (9).

In the last years monoclonal antibodies (MoAb) have been wide used in the treatment of different cancers. Since myeloblasts are present in the blood and the bone marrow they are a suitable target for MoAb. MoAb generally function by binding on cell surface antigen and then signaling the cell to the immune system kills the malignant cell. The use of MoAb is limited by two principal issues, the first most of the suitable antigens are also expressed by healthy immune cells. The second, MoAb have shown to be ineffective in AML if they are not conjugated with toxic agents. MoAb anti CD33 have been developed (which is expressed in the 90% of leukemia cells). One of the strategies is to link the anti-CD33 MoAb with cytotoxic agent (for example calicheamicin, a molecule that is determining damage to the DNA). The principle in this case is that the MoAb is bounded to the surface of the myeloblast and once it is internalized the cytotoxic agent kills the cell. Another target for MoAb is CD123, a gene expressed in the vast majority of AML patients. There is also a new approach in MoAb therapies, such as the MoAb designed to cross-link myeloblast to immune cells (typically T-cells or NK-cells). This new approach is showing promising results and is in clinical evaluation (10).

A different approach in MoAb is directed to reactivate immune system against myeloblast instead of the directly killing the myeloblast. The immune system is conducting immune surveillance against mutated cells. Malignancies occur when the cancer cells are able to escape the immune system. Often this happens suppressing the immune response against cancer. In the last years, a growing interest is directed on what are called immune checkpoints. Immune checkpoints are proteins that are able to regulate the immune response, negative immune checkpoints are dampening the immune response while positive checkpoints are promoting immune response. In different type of malignancies, MoAb against negative immune checkpoints are entering in clinics showing very promising results. As examples there are MoAb directed against CTLA4, PD-1, CD47 (10).

Chimeric Antigen Receptor (CAR) T-Cell Therapy is a different approach exploiting the immune system against the tumor. CAR T-cells are genetically engineered cells to express particular molecules on the surface to exert tumor-lytic activity. In many different cancers the CAR T-Cell Therapy has shown appreciable results, leading to the idea of developing a similar approach to AML (10).

The study of the biology of AML is leading to a better comprehension of the disease. This increasing knowledge has led researchers to identify new targets in AML and it has been followed by the development of new molecules. Indeed, there are a lot of ongoing clinical trials that are evaluating the efficacy of these molecules and approaches in AML. A challenge for clinicians is to choose the optimum treatment for the patient, with a particular attention to elderly or unfit patients. Clinicians should choose the optimal treatment according to age, comorbidities and the genetic mutations. This requires genome analysis after the diagnosis to choose the optimal treatment. However, after the treatment AML can relapse and present new mutations, then an accurate analysis is necessary to choose the optimal treatment. Deep sequencing analysis have shown that myeloblasts are heterogenous and a percentage of the primary tumor presents different alterations. Moreover, this subclone can expand and acquire other mutations. These new molecules that can be used to treat relapse cases will lead to new acquired resistance (as new gene mutation, different alteration in the signaling pathways and so on) requiring to study alternative strategies. For example, AML patients treated with FLT3 or IDH2 inhibitors have already developed resistance (different studies showed different mechanism of acquired resistance to FLT3 inhibitors, leading the need to ulterior strategies) (11).

Gene mutations and genomic alteration in AML are the basis of the pathogenesis and the basis for the relapse. Deep sequencing of AML has permitted to better classify AML in different subgroups. Moreover, the mutation profiles allowed to stratify patients in risk group and to better predict prognosis. They have also permitted the development of new therapies. These more specific therapies proved to be more tolerable in older patients which cannot be treated with standard chemotherapy. However, this is not enough, the genetic alterations present in AML is beyond the markers we highlighted before. Moreover, some of these mutations arise at different state of disease (primary, relapse, after treatment), meaning that our stratification based on these genetic abnormalities is still inaccurate. The discovery of new mutations or other alteration (like in signaling pathway) in primary tumors or in relapses would allow the development of new primary therapies or for secondary line. Another important aspect of these gene mutation is the possibility to early discovery relapse trough molecular analysis. There is also interest in the epigenetic landscape which could lead to other promising therapy and can be used as an early marker for relapse. At the current state, we have less knowledge on the epigenetic landscape in AML during progression and it’s role in relapse and resistance to therapy. At the same time immunotherapy is still in development, notwithstanding it is showing promising results there are still some black spot we have to address. New therapies or combination of them can lead malignancy eradication during the preleukemic stage or before relapse. Another challenge that has to be address is marker identification of leukemia stem cells which are remaining after therapy and prompting the relapse. The gain of knowledge in these fields is the basis for personalized medicine. It is expected that the right combination of these therapies will be choose according to the specific subgroup of patients. Thus, there is need of new preclinical study to identify the right subgroup for the right combination of therapies. At the same time, it is also necessary to improve trial clinical design.

An open question is how myeloblasts remodel the bone marrow to create a permissive niche to sustain tumor growth. In the same way, AML is leading to metastasis where malignant cells interact with different type of cells. These processes involve many different molecules involved in adhesion, proliferation and differentiation. To address these last questions, we need to understand interactions with other cells that are presents in the bone marrow.

In conclusion, there are still many open questions that have to be addressed by researchers and as discussed in the next articles there is room for machine learning.

following article on medical image analysis with machine learning: here.

following article on machine learning using other data source: here

Selected bibliography:

1. Pievani A, Biondi M, Tomasoni C, Biondi A, Serafini M. Location First: Targeting Acute Myeloid Leukemia Within Its Niche. J Clin Med [Internet]. 2020 [cited 2020 Sep 29];9. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7290711/

2. Acute myeloid leukemia [Internet]. Wikipedia. 2020 [cited 2020 Oct 8]. Available from: https://en.wikipedia.org/w/index.php?title=Acute_myeloid_leukemia&oldid=976980044

3. Fatonah NS, Tjandrasa H, Fatichah C. Automatic Leukemia Cell Counting using Iterative Distance Transform for Convex Sets. International Journal of Electrical and Computer Engineering (IJECE). 2018;8:1731–40.

4. Hwang SM. Classification of acute myeloid leukemia. Blood Res. 2020;55:S1–4.

5. Appelbaum FR, Gundacker H, Head DR, Slovak ML, Willman CL, Godwin JE, et al. Age and acute myeloid leukemia. Blood. 2006;107:3481–5.

6. Papaemmanuil E, Gerstung M, Bullinger L, Gaidzik VI, Paschka P, Roberts ND, et al. Genomic Classification and Prognosis in Acute Myeloid Leukemia. N Engl J Med. 2016;374:2209–21.

7. Genomic and Epigenomic Landscapes of Adult De Novo Acute Myeloid Leukemia. N Engl J Med. 2013;368:2059–74.

8. Kirtonia A, Pandya G, Sethi G, Pandey AK, Das BC, Garg M. A comprehensive review of genetic alterations and molecular targeted therapies for the implementation of personalized medicine in acute myeloid leukemia. J Mol Med. 2020;98:1069–91.

9. Hou H-A, Tien H-F. Genomic landscape in acute myeloid leukemia and its implications in risk classification and targeted therapies. J Biomed Sci [Internet]. 2020 [cited 2020 Oct 7];27. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7372828/

10. Miyamoto K, Minami Y. Cutting Edge Molecular Therapy for Acute Myeloid Leukemia. Int J Mol Sci [Internet]. 2020 [cited 2020 Sep 29];21. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7404220/

11. Green SD, Konig H. Treatment of Acute Myeloid Leukemia in the Era of Genomics — Achievements and Persisting Challenges. Front Genet [Internet]. 2020 [cited 2020 Sep 29];11. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7267060/

Peter Moss Leukemia AI Research

Free & Open-Source Technologies for the fight again Leukemia.

Salvatore Raieli

Written by

Salvatore is currently working as bioinformatician researcher. During is PhD in immunology he became passionated about bioinformatics, machine learning and AI

Peter Moss Leukemia AI Research

Researching into AI & modern technologies, and how they can be used in the fight against Leukemia.

Salvatore Raieli

Written by

Salvatore is currently working as bioinformatician researcher. During is PhD in immunology he became passionated about bioinformatics, machine learning and AI

Peter Moss Leukemia AI Research

Researching into AI & modern technologies, and how they can be used in the fight against Leukemia.

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store