Selection by chemotherapy or CAR-T for malignant B cells containing inactivating mutations and/or deletions of key apoptosis/cell death genes, e.g., TP53 (encoding p53), CDKN2A (encoding p16), and PMAIP1 (encoding NOXA), exemplifies Darwinian processes in resistance onset and propagation (1) (Figure 1). However, coordinated downregulation of CD19 and CD22, simultaneous with BCR and BTK preservation or upregulation, seems an unlikely consequence of random or accidental genetic events. That is, an exclusively Darwinian model, especially over the short time-scales observed experimentally by Aminov et al., seems unlikely, and another process must explain how these phenotype changes emerge and stabilize. CD19 is normally activated upon hematopoietic stem cell commitment into the B cell lineage, and then progressively increases in expression with onward B cell lineage maturation. It functions as a coreceptor or accessory to the BCR/BTK pathway, signaling into B cell lineage cells via the PI3K/AKT pathway, and in this way contributes to B-lineage differentiation and specialization (7–9). Given this normal function of CD19, insofar as malignant B cells survive weapon payloads attached to CD19-targeting treatments, the CD19 targeting itself will affect the cellular phenotype, either by the direct inhibition of CD19 function, or indirectly through the selection of cells on the lower end of the CD19 expression spectrum (Figure 1). The state of having less CD19 function and expression can be expected to favor maturation arrest early in the B cell lineage differentiation continuum, when CD19 and CD22 expression is lower but BCR and BTK expression is preserved (Figure 1). That this is the case is also supported by the observation by Aminov et al. that the master transcription factor (MTF) SOX4 was upregulated in the resistant cells (Figure 1). Viewed in this light, other predictions can be made, e.g., CD10 could be a potential alternative to CD22 as a surface target for salvage therapy — CD10-targeting CAR-T is being explored (10) (Figure 1).

Phenotype of B cell malignancies at diagnosis and resistance depends on the expression of B cell lineage MTFs during different stages of B cell lineage differentiation. (A) In B cell malignancies, treatment resistance occurs via two evolutionary models: A Darwinian model for treatment resistance occurs when a preexisting mutation confers resistance and is selected for during treatment, e.g., selection by chemotherapy for TP53 mutations and/or deletions. Alternatively, a Lamarckian model predicts that an oncotherapeutic can directly trigger adaptive responses in malignant cells. The baseline MTF configuration of the cells constrains the range of adaptive shifts, which are propagated to daughter cells via epigenetic mechanisms. Notably, both models can act concurrently. (B) Coordinated shifts in the expression of lineage MTFs and surface receptors occur during normal B cell lineage differentiation. Treatments that target CD19 or BTK affect inputs into the lineage MTF circuit and may alter malignant phenotypes in a pattern dictated by the baseline MTF configuration and B cell lineage differentiation. (C) CD10 and CD19 are expressed at different stages in B cell lineage differentiation and present targeting opportunities. B cells undergoing differentiation that have been targeted by CD19 treatments may shift phenotypes to confer resistance but remain susceptible to alternative targeted therapies such as anti-CD10.

Lamarck proposed directed evolution such that the environment instructs pro-fitness modifications (aka adaptation) heritable by subsequent generations (11) (Figure 1). Although it is difficult to transmit adaptation occurring at a somatic level into a separate germline compartment, unicellular neoplastic evolution faces no such difficulties because adaptive shifts in gene expression are readily propagated to daughter cells via epigenetic inheritance. Moreover, persistence of the environmental cue, e.g., CD19 CAR-T, can be expected to reinforce the adaptations in daughter cells, since these shifts after all emerge organically and predictably from cell physiology networks. Aminov, Giricz, and colleagues found that the suppression of CD19 and CD22 was indeed an epigenetic, Lamarckian process and not genetic (3). There is, thus, a further therapeutic implication: epigenetic enzymes mediating repression of CD19 and CD22 also constitute candidate targets for salvage therapy.