Stem cell biology was a boldly novel discipline in the early 20th century, beginning with the proof of existence of bone marrow (BM) hematopoietic stem cells (HSCs) by James Till and Ernest McCulloch in the 1960s after successfully reconstituting blood cells on heavily irradiated mice [1]. On a societal level, at the time, Till and McCulloch’s discovery also contributed to the theoretical framework supporting the clinical implementation of BM transplantation for those exposed to severe nuclear radiation during the Cold War [2]. In the decades that followed, expansion of cellular and molecular characterization of the physiological properties of HSCs scientifically and militarily led to the establishment of an intricate blood cell differentiation hierarchy [3] and, more importantly, the development of technologies such as fluorescence-activated cell sorting [4] and advances in tissue culture assays, including the long-term culture-initiating-cell (LTC-IC) assay that allows the differential clonal output potential of single HSCs to be examined; these all accelerated HSC research [3,5].

About 30 years after the discovery of HSCs, in the context of oncogenesis, the concept of the existence of a select subpopulation of leukemic cells not only capable of sustained self-renewal but also able to repopulate and give rise to human leukemias was demonstrated in vivo in acute myeloid leukemia (AML) by John Dick and colleagues [6,7], as well as in chronic myeloid leukemia (CML) by Connie Eaves and colleagues [8,9,10]. These cells with strong leukemia-initiating capacity, termed leukemic stem cells (LSCs), are found to be immunophenotypically enriched within the Lin⁻CD34⁺CD38⁻ BM cell fraction, despite the high degree of surface and intracellular marker heterogeneity therein [11,12,13]. The experimental basis for such seminal discoveries involved using limiting dilution of fractionated cell pools harboring distinct surface markers, followed by engraftment into immunocompromised murine hosts. These techniques, though time-consuming, effectively isolated cell population enriched for LSCs, allowed relative quantification of LSC frequency, and even served as a contemporary in vivo gold-standard approach for functional validation of LSC activity [14]. With the advent of next-generation sequencing, high-throughput multi-omics technologies, and sophisticated murine models developed in the 21st century, scientists are now able to further interrogate the nature and evolutionary trajectory of LSCs, sprouting the seed for an era of advanced stem cell research. Indeed, recent mounting evidence corroborates that LSCs are heavily involved in refractory hematological malignancies, particularly AML and CML. 

Despite recent advancements made in characterizing LSCs, it is worth noting that LSCs should not be perceived to have well-defined, uniform biomarkers. Rather, LSC surface markers can be fluidic and even context-dependent. Dick and colleagues’ seminal discovery in 1997 uncovered that cells expressing surface markers CD34⁺CD38⁻ were able to differentiate in vivo in severe-combined-immunodeficiency-disease (SCID) mice into leukemic AML blasts; however, this description was broad and preliminary, as studies later revealed this same set of surface marker is also shared by normal HSCs and progenitor cells, sparking interest in the search of cellular markers that specifically encapsulate LSCs [15,16,17]. Moreover, early characterization of AML LSCs was predominantly antibody-based; however, such practices may lead to biases resulting from differing reactivities and specificities of the antibodies used and the harsh assay environment associated with cell sorting/fractionation, as not all cell types could be sorted with the same efficiency in vitro and in vivo [18,19]. Nowadays, recognizing these potential pitfalls, research efforts have concentrated on examining surface antigens that appear to be aberrantly regulated or differentially expressed on LSCs as opposed to their healthy stem and progenitor cell counterparts. One promising marker is CD33, which is highly expressed in AML patients, and its expression along with T cell immunoglobulin and mucin protein (TIM3) specifically denotes AML cells as opposed to normal hematopoietic tissues [20,21]. Another marker that is reported to be preferentially overexpressed in CD34⁺CD38⁻ AML cells is CD123, and its expression serves as a clinical marker for adverse patient outcome; it is potentially mediated by mechanisms involving STAT5 activation [22,23,24,25]. Furthermore, CD47, which upon interaction with SIRPα on circulating macrophages and dendritic cells inhibits phagocytosis to facilitate immune evasion of AML cells, is overexpressed in CD34⁺CD38⁻ LSCs, conferring a survival advantage to AML stem cells [26,27,28]. Using signal-sequence-trap technology to identify surface antigen expression milieu, followed by quantitative real-time polymerase chain reaction (qRT-PCR) validation across Lin⁻CD34⁺CD38⁻ AML samples and normal BM-derived HSC/progenitor cells, Hosen and coworkers found CD96 to be preferentially upregulated in AML cells and demonstrated LSC activity among CD34⁺CD38⁻CD96⁺ AML cells through successful engraftment of these cells into irradiated immunocompromised mice [29]. Furthermore, transcriptomic analysis done on purified HSCs from myelodysplastic syndrome (MDS) patients and age-matched cord blood cell HSC controls identified selective overexpression of CD99 on LSCs, and that CD99 is at a particularly high level at relapse when compared to samples obtained at time of diagnosis, suggesting a potential link between CD99 expression and the chemo-resistant properties of LSCs in general [30]. Furthermore, a study conducted by Heo and colleagues reported prominent enrichment of the CD45^dimCD34⁺CD38⁻CD133⁺ cells in BM of AML patients, and that this unique signature is associated with poor prognosis; however, functional validation of LSC activities of these cells is still needed [31]. Another factor that complicates the already heterogenous landscape of AML LSC surface antigen expression is the stage during which leukemic transformation occurs, as recent evidence reported transformation may also take place in mature granulocyte–macrophage precursors without expressing CD34, specifically in those samples with NPM1 mutation [32,33,34]. In other instances, leukemic-initiating cells are found to exist in the CD38⁺ progenitor cell population [35]. These novel insights may challenge the way in which we purify and interpret the surface antigen milieu of AML LSCs. In a nutshell, AML LSCs harbor an extremely complex surface antigen expression profile, and further consensus is needed to define these molecular markers. Of equal importance is the mechanistic understanding underlying such surface markers in relation to stemness and clinical drug resistance.

The first evidence of a highly primitive type of leukemic cells present in CML patients was reported in 1992 using techniques such as LTC-IC assay followed by limiting dilution assays of patient CML blood and BM samples [36,37]. A later study published in 1999 from the same research group established the quiescent nature of CML LSCs after isolating a quiescent cell fraction among CD34⁺ CML LSCs using nucleic-acid-binding agents Hoechst 33342 and Pyronin Y, and subsequently confirming the engraftment capability of these quiescent CML LSCs on immunodeficient mice [38]. This dormant state was later discovered to be an important feature of drug-resistant CML LSCs, as they contribute to the root cause of drug resistance and disease relapse in CML patients [39,40]. Since then, many studies have aimed to characterize surface biomarkers selectively present on CML LSCs; however, just like the dilemmas encountered in defining the surface marker landscape of AML LSCs, evidence in CML LSCs appears to be controversial due, at least in part, to the lack of standardized in vitro and in vivo techniques employed and the molecular heterogeneity intrinsic to the functioning of CML LSCs throughout different stages of disease progression [41]. Nevertheless, studies to date have mainly focused on examining the surface marker expression of CD34⁺CD38⁻ CML cells in the chronic phase, a comparatively stable stage of CML pathogenesis. According to Herrmann and coworkers, CD34⁺CD38⁻ CML patient cells in the chronic phase are reported to have a roughly 10-fold higher expression of CD33 compared to normal CD34⁺CD38⁻ stem cells by flow cytometric analysis, even though CD33 remains comparable at the transcript level [42]. Interestingly, IL-1 receptor accessory protein (IL1RAP), which is found to be upregulated in CD34⁺ and CD34⁺CD38⁻ CML cells, appears to increase as patients transition from the chronic phase to accelerated and blast crisis phases [43]. Furthermore, using gene array, qPCR, and flow cytometric analyses, Hermann et al. reported that CD34⁺CD38⁻CD26⁺ but not CD34⁺CD38⁻CD26⁻ LSCs obtained from chronic-phase CML patients contained BCR/ABL1 mRNA, exhibited repopulating capability in NSG mice, and were highly expressed in imatinib-nonresponder patients, suggesting that CD26 is a highly specific CML LSC biomarker and a potential therapeutic target [44,45]. Complementing this line of evidence, a more recent study using single-cell gene expression analysis in conjunction with immunophenotypic screening revealed a distinct signature of Lin-CD34⁺CD38^−/lowCD45RA⁻cKIT⁻CD26⁺ as being associated with a particular CML LSC compartment that is relatively insensitive to tyrosine kinase inhibitor (TKI) treatment [46]. Furthermore, a study published in 2020 comparing a total of 878 BM or blood samples from 274 patients with AML, 97 patients with CML, and 288 controls also revealed an aberrant profile of CD25⁺CD26⁺/CD56⁺/CD93⁺/IL1RAP⁺ antigens among CD34⁺CD38⁻ CML cells [11]. Indeed, characterization of the surface marker landscape of CML LSCs not only aids in the phenotypic purification of these intriguing cells, but also paves the way for the identification of specific, actionable, and prognostic biomarkers on CML LSCs for therapeutic design and prevention of disease relapses. As exciting as these discoveries may seem, further studies are still needed to identify CML LSC surface antigens potentially driving disease progression from chronic to accelerated and blast crisis phases and/or conferring TKI resistance.

Owing to their unique molecular properties, including self-renewal, phenotypic plasticity, and the wondrous ability to differentiate and repopulate new tissues, LSCs are highly versatile entities, insidiously perpetuating therapeutic resistance [47,48]. Indeed, genetic variants known to evade therapeutic treatments are enriched in stem-like cells found across almost all forms of leukemia [49,50,51,52,53]. Among the molecular aberrations present in cancer stem cells, drug-induced senescence and quiescence represent predominant causes of patient relapse, as most antineoplastics target processes associated with actively dividing bulk cancer cells, not those in the quiescent state [54,55]. As LSCs represent just a rare fraction of the total bulk cancer cells, patients undergoing chemotherapy can be considered to have reached therapeutic endpoints when, in fact, those LSCs may still remain alive. More so, several groups have also reported that cancer stem cells can become “trained” to respond to chemotherapeutic insults either through crosstalk with the tumor microenvironment or acquisition of adaptive genomic circuitry to mitigate drug-mediated cytotoxicity or to bypass drug-targeted pathways, fortifying their ability to withstand therapeutic challenges [56,57,58]. This creates a clinical dilemma where drug resistance and relapses are not only permitted, but are sometimes even inadvertently encouraged once patients go through round after round of therapy. These complications, coupled with the heterogenous surface marker expression of LSCs, nonspecific cytotoxicity of existing antineoplastics, and inter-patient sensitivity to standardized treatments, collectively stagnate the overall survival of leukemia patients, especially that of elderly individuals.

2. Surface Antigen-Based Immunotherapies

3. Small-Molecule Inhibitors

4. Combination Therapies