JAM-C Expression as a Biomarker to Predict Outcome of Patients with Acute Myeloid Leukemia—Response

Acute myeloid leukemia (AML) is a heterogenous disease affecting essentially patients older than 55. Although most patients experience complete remission after intensive chemotherapy, the disease remains of dismal prognosis with long-term overall survival (OS) below 40% (1). Therefore, there is a need for the identification of additional biomarkers for a better stratification of patients with AML. We are pleased to see that JAM-C (JAM3) expression, which we have associated to leukemia-initiating cell (LIC) activity and bad prognosis in de novo AML (2), has been investigated in a large independent cohort of patients with AML (3). In their study, Von Bonin and colleagues measured the JAM-C expression by flow cytometry on bulk leukemic cells of 198 AML samples including 2% of secondary AML and 11% of therapy-related AML. In agreement with our published results, they found that JAM-C expression was associated with a significantly reduced OS. They also showed that JAM-C expression was inversely correlated with CD34 expression and did not impact the relapse-free survival (RFS). These last results are in apparent contradiction with our findings that JAM-C identifies a pool of LIC and that the frequency of JAM-C–expressing LIC is associated with reduced leukemia-free survival (LFS). However, these differences may be due to differences between our respective flow cytometry gating strategies, sample origin (bone marrow vs. blood), or clinical parameters of patient samples, which we discuss hereafter.

The valuable finding of Von Bonin and colleagues that a cutoff of 10% JAM-C⁺ cells in the leukemic bulk has a prognostic value and predicts lower probability to achieve complete remission confirms our previous results, despite the differences in the percentage of JAM-C⁺ cells between the two studies. This may be due to differences in flow cytometer settings, in antibody panels or in flow cytometry gating strategies. Such differences could be easily corrected through normalization of flow cytometry panels and instrument settings to standardize the use of JAM-C as a new predictive prognostic marker. Another possibility is that the threshold value based on 10% routine cutoff differs from our threshold calculation made by statistical curve fitting of “JAM-C–expressing cell frequencies” within the subset of the most immature leukemic cells defined as CD45^dimCD33⁺CD34⁺CD38^lowCD123⁺CD41⁻. Such a flow cytometry gating strategy allowed us to eliminate positive events due to JAM-C expression by normal hematopoietic stem cells or platelet sticking on leukemic blasts. This may explain the differences between the two studies in term of impact on RFS and correlation with CD34 expression.

Indeed, the inverse correlation between JAM-C and CD34 expression found by Von Bonin and colleagues is counterintuitive for a biomarker associated with LIC activity. Nevertheless, several studies have demonstrated that LICs are not restricted to CD34⁺ cells and that LIC activity may even be enriched in the CD34⁻ compartment for AML, having less than 10% of CD34⁺ cells (4, 5). Our results have thus been reanalyzed for correlation between frequencies of JAM-C–expressing cells and the CD34⁺CD38^low compartment. We found an inverse correlation between the frequencies of CD34⁺CD38^low cells and SSC^loCD45^dimCD33⁺CD34⁺CD38^dimCD123⁺JAM-C⁺ cells, reinforcing the idea that JAM-C identifies LIC within the CD34⁺CD38^low compartment even for samples with less than 10% of CD34⁺ cells (Fig. 1). Alternatively, it may well be that JAM-C expression by leukemic subsets plays a role in drug resistance, irrespective of the maturation compartment of leukemic cells. This would be in agreement with a cell autonomous function of JAM-C in leukemic cells as reported recently (6). Indeed, JAM-C has been shown to associate in cis with LRP5 and to signal through AKT/β-catenin/CCND1 axis to maintain leukemic cell self-renewal. Although it is not known whether such a molecular mechanism may account for drug resistance of bulk leukemic cell subsets, results already indicate that JAM-C is a robust biomarker of poor disease outcome, irrespective of the gating strategy or antibody reagent used to measure JAM-C expression.

Overall, the use of JAM-C as a biomarker on the leukemic AML bulk and its use for identification of LIC in the CD34⁺ compartment are highly complementary. JAM-C as a biomarker is helpful for patient stratification and decision-making, whereas measurement of JAM-C⁺ cells endowed with LIC activity may provide new methods to follow minimal residual disease.

See the original Letter to the Editor, p. 6339