The protocol of this systematic review was registered a priori in the International Prospective Register of Systematic Reviews (PROSPERO) with registration number CRD42023481493.22

The search for articles was carried out by two reviewers, on September 16th, 2023, in six bibliographic databases: PubMed, Embase, Scopus, Web of Science, the Cochrane Library, and SciELO. The terms used were “acute myeloid leukemia” and all of its synonyms in MeSH Terms (or Emtree, in Embase) AND ‘CD36’ and all of its synonyms in MeSH Terms (or Emtree, in Embase). The complete search strategy can be accessed in the PROSPERO protocol.22 Additional research was also carried out using Google Scholar before extracting data, looking for new studies not yet peer-reviewed.

Studies that analyzed the expression of CD36 in human patients, including clinical trials, case-control studies, and case reports were included. The articles should evaluate the impact of CD36 on factors related to the prognosis of AML, including, for example, survival and response to treatment. Reviews, editorials, letters, abstracts from conference annals, and studies that only used animals or lineage cells for the analyses of CD36 were excluded. Studies that addressed only one specific subtype of AML and not the disease spectrum were also excluded.

Two independent reviewers selected studies based on eligibility criteria. Disagreements between individual judgments were solved by a third reviewer. The software used for blinding and recording decisions was Rayyan.23

From each eligible study, data were input in duplicate in an Excel worksheet, following a standardized template created for this review. The following data were extracted from articles: first author, publication year, research location, study design, sample size, age of participants, percentage of males, eligibility criteria, stem cell source, method of CD36 quantification, comparators, statistical analysis, method of AML classification, evaluated prognostic factors and results related to the impact of CD36 on factors influencing the prognosis of AML.

The methodological quality of nonrandomized studies was evaluated using the Newcastle-Ottawa Scale. Studies rated 3–4 stars in the selection domain, 1–2 stars in the comparability domain and 2–3 stars in the outcome/exposure domain were classified as good quality. Studies rated 2 stars in the selection domain, 1–2 stars in the comparability domain and 2–3 stars in the outcome/exposure domain were considered as fair quality. Those rated 0–1 star in the selection domain or 0 stars in the comparability domain or 0–1 star in the outcome/exposure domain were classified as poor quality.24 Two independent reviewers evaluated the quality of the studies, and disagreements between individual judgements were solved by a third reviewer to reduce the risk of bias.

The search strategy resulted in 905 articles, and one more article was added after a manual review to identify eligible articles that might not have been captured by the search strategy. After excluding 305 duplicates, the titles and abstracts of 600 articles remained were assessed and another 583 articles were excluded based on the eligibility criteria, leaving 17 articles and one added manually giving a total of 18 to be read in full. Of these, nine articles25–33 were included in this review with the other nine being excluded for not complying with the inclusion criteria (six did not report the prognoses, two were conference abstracts and one was exclusively an animal study). The agreement between evaluators during the full-text phase was 93.75 % (Cohen’s κ: 0.875). Figure 1 represents the flowchart detailing the excluded articles at each stage, along with the reasons for their exclusion.

Figure 1.

Flow diagram of study selection process.

Table 1 summarizes the characteristics of the studies included in the main analysis. Articles were published between 1992 and 2023. Some of the articles do not provide all the collected data, such as the mean age and gender of the study participants. From the studies that reported the mean age of the participants, the mean age ranged from more than one month to approximately 44 years. The stem cell source in most articles was the bone marrow. Two studies used peripheral blood in addition to bone marrow, and one of the articles used databases and did not clarify the stem cell source but mentioned the use of tumor tissue. The most commonly used method for CD36 quantification was flow cytometry, in addition to RNA sequencing and immunohistochemistry. The comparison was mainly conducted between patients with complete remission of AML versus patients who relapsed, patients with positive or increased CD36 expression versus negative or decreased CD36 expression, but there were also comparisons with healthy individuals and people with other types of cancer. The articles primarily classified AML based on the French-American-British (FAB) classification and Cytogenetic and Molecular Risk Groups. The studies demonstrate the impact of CD36 on the prognosis of AML across different categories, such as survival, remission, chemotherapy resistance, and tumor cell proliferation. All results linking CD36 expression to any of these characteristics were collected.

The methodological quality of the studies analyzed in this review ranged from low to good, with most of the articles being classified as fair quality, primarily due to significant gaps in crucial information reported by the articles. Scores on the Newcastle-Ottawa Scale for the studies ranged from 5–7 stars.

In a prospective study investigating the prognostic significance of cell surface antigens associated with myeloid differentiation, such as CD36, the authors found that the presence of these myeloid-associated cell surface antigens did not have prognostic significance for children.25 However, a higher frequency of CD36-positive cells at diagnosis was identified in cases of children who experienced disease recurrence.26 In contrast, in adults, CD36 expression was significantly higher in AML when compared to other types of cancer,27 and the increase or even the expression of CD36 is associated with several factors that appear to be associated with a poor prognosis. Bone marrow mononuclear cells from AML patients expressing CD36 are less susceptible to chemotherapeutic agents such as cytosine arabinoside (Ara-C) compared to cells that do not express CD36,28 demonstrating that CD36 expression interferes with chemotherapy. Another study also demonstrated a tendency for a lower response to chemotherapy in the group of patients with positive CD36 expression compared to the group with negative CD36 expression.29

Furthermore, CD36 expression is associated with relapse and recurrence of the disease. The higher frequency of CD36-positive cells at diagnosis was identified in cases that presented recurrence of the disease,26 as well as a higher risk of relapse for CD36+ patients, thereby presenting a shorter leukemia-free survival rate.30

In addition to a higher risk of relapse, CD36 expression is also associated with shorter overall survival. Five-year non-survivors showed increased levels of CD36-positive AML blasts,31 while complete remission was achieved in a higher percentage of CD36-nagative patients, as well as overall survival rates and leukemia-free survival rates.29

CD36 expression in blasts at diagnosis is also associated with human AML progression and relapse. CD36 was significantly associated with worse survival. The survival of CD36-high patients was half that of CD36-low patients. Furthermore, a high CD36 protein expression at diagnosis was associated with an increased cumulative incidence of relapse after intensive chemotherapy, and a multivariate analysis showed that a high expression of CD36 was associated with a shorter cumulative incidence of relapse.32

Some proteins may be associated with this CD36 effect, such as apolipoprotein C-II (APOC2), which cooperates with CD36 to promote leukemia growth, as described by Zhang et al.33 Using a dataset of 23 patients, high CD36 expression had shorter, but not statistically significant, overall survival. Nevertheless, analyzing a dataset of 173 patients, those with high APOC2 or CD36 levels had significantly shorter overall survival than patients with low APOC2 and low CD36 expressions.33

A connection between CD36 expression and the tumor immune microenvironment is suggested. CD36 expression is significantly positively correlated with infiltrating stromal scores in AML, and there are positive correlations between CD36 expression and infiltrating levels of immune score in AML, as well as infiltrating levels of the Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE) score in AML.27 Additionally, negative correlations were observed between CD36 expression and four methyltransferases in AML.27

CD36 expression was more frequently observed in an unfavorable cytogenetic group,29 suggesting a link between CD36 and high-risk cytogenetic profiles. Cytogenetic and molecular abnormalities appear to contribute to the severity driven by CD36 expression. CD36-positive patients with trisomy 8 had significantly poorer leukemia-free survival compared to CD36-positive patients without trisomy 8,30 highlighting the interplay between CD36 and specific cytogenetic alterations. Supporting this association, intermediate and unfavorable karyotypes exhibited a higher proportion of CD36-expressing blasts.32 CD36 expression was characterized as an independent marker for AML progression, but has a higher association in patients with FLT3 abnormalities, reinforcing its role in poor prognosis.32 Moreover, AML cases in the CD36-high group were enriched with KMT2A (11q23) or t(9;22) abnormalities, while cases in the CD36-low group showed a higher frequency of favorable cytogenetic abnormalities, such as t(15;17) and t(8;21).32 This suggests that CD36 expression correlates with more aggressive disease phenotypes driven by adverse genetic features.