Is ctDNA the Road Map to the Landscape of the Clonal Mutational Evolution in Drug Resistance? Lessons from the PALOMA-3 Study and Implications for Precision Medicine

Preclinical studies have established the role of the cyclin D1–cyclin-dependent kinase 4 and 6 (CDK4/6)–retinoblastoma (RB) pathway in the normal development of the mammary gland luminal compartment and in luminal-type breast cancers (1). CDK4/6 phosphorylates RB, releasing the inhibition on the E2F transcription factor to promote expression of CCNE1 (the gene that encodes cyclin E). This activates the cyclin E–CDK2 complex and allows progression to the S phase of the cell cycle. The activity of the cyclin D1–CDK4/6 complex is opposed by members of the INK4 CDK inhibitor family, of which p16^INK4a is the founding member. Using a large panel of human breast cancer cell lines, Finn and colleagues showed that a selective CDK4/6 inhibitor preferentially inhibits the proliferation of estrogen receptor–positive (ER⁺) cell lines (2). At the molecular level, the potent activity of the CDK4/6 inhibitors in ER⁺ breast cancers is linked to the direct transcriptional regulation of CCDN1 by ER and to the integrity of the cyclin D1–CDK4/6–RB pathway, which is retained in the majority of ER⁺ breast cancers (3, 4).

A number of phase III studies have shown a substantial improvement in progression-free survival (PFS) with the addition of the CDK4/6 inhibitors in metastatic ER⁺, HER2-negative breast cancers, reviewed in ref. 5. Currently, selective CDK4/6 inhibitors, including palbociclib, ribociclib, and abemaciclib, in combination with endocrine treatment, are the standard of care in advanced ER⁺, HER2-negative breast cancer. These drugs have some degree of difference in their selectivity against various CDKs and toxicity profiles. Although the CDK4/6 inhibitors are highly effective, not all patients benefit from CDK4/6 inhibitors, and the majority of patients ultimately develop resistance. Preclinical studies have identified several putative mechanisms of resistance, including loss of RB function (e.g., by RB1 mutations), CCNE1 amplifications or overexpression, CDK6 amplifications, higher levels of p16^INK4a, and increased growth factor signaling, particularly the PI3K–MTOR signaling pathway, which cross-talks at multiple levels with CDK4/6 (Fig. 1A), reviewed in ref. 5. However, the clinical intrinsic and acquired mechanisms of resistance and predictive biomarkers of response remain largely elusive. One of the challenges of studying these mechanisms is the need for sequential sampling of the metastatic disease. To overcome this challenge, liquid biopsies from plasma cell-free DNA to detect circulating tumor DNA (ctDNA) are emerging as a tool with high sensitivity.

In this issue, O'Leary and colleagues performed a retrospective analysis of ctDNA obtained from a subset of patients from the PALOMA-3 clinical trial to study mechanisms of acquired resistance to palbociclib and fulvestrant (6). The PALOMA-3 study was a phase III clinical trial that randomized patients with ER⁺, HER2-negative metastatic breast cancer that had previously progressed on an endocrine treatment to fulvestrant plus palbociclib or fulvestrant plus placebo (Fig. 1B). This study showed significant improvement in median PFS in the combination arm (7). Earlier studies used tissue and liquid biopsies from the PALOMA-3 trial to examine intrinsic resistance and early ctDNA dynamics as predictors of clinical benefit. These studies indicated that day 1 (pretreatment) ctDNA PIK3CA and ESR1 mutations and the level of ER expression were not associated with palbociclib treatment effect. In contrast, higher expression of CCNE1 mRNA was associated with relative resistance to palbociclib. In addition, early changes in PIK3CA ctDNA dynamics predicted PFS on palbociclib and fulvestrant more strongly than ESR1 dynamics (8, 9).

To investigate mechanisms of acquired resistance to palbociclib plus fulvestrant versus fulvestrant alone, O'Leary and colleagues conducted whole-exome sequencing (WES) and targeted sequencing of paired day 1 and end-of-treatment (EOT) ctDNA obtained from the PALOMA-3 trial patients (6). This analysis showed that clonal evolution resulting in the acquisition of driver mutations commonly occurs during treatment with fulvestrant alone or fulvestrant and palbociclib. Furthermore, the presence of these acquired mutations was associated with longer PFS, suggesting that early (intrinsic resistance) and late (acquired resistance) progression have distinct mechanisms of resistance. The authors suggest that the emergence of these clones is either due to mutational processes, as evident by the presence of the APOBEC and mismatch-repair signatures, or due to the selection of preexisting rare subclones. Importantly, WES data performed on a limited number (n = 14) of day 1 ctDNA revealed several mutations potentially relevant to the development of endocrine resistance beyond the ESR1 mutations (mutations in the NOTCH family and NF1). Thus, a broader discovery approach to identify private genetic aberrations, which in part may converge to a common pathway, is likely needed to identify potential drivers of resistance in a given tumor. These genetic aberrations may also be involved in intrinsic resistance to palbociclib and fulvestrant, inferring the next line of treatment.

The sequencing analysis also revealed that Rb1 mutations were the only mutations exclusively acquired in the palbociclib and fulvestrant treatment arm and not the fulvestrant-only arm. However, despite previous preclinical and clinical evidence that RB function is required for the efficacy of the CDK4/6 inhibitors, in the PALOMA-3 study RB1 mutations were found to be acquired in only 5% of patients through palbociclib and fulvestrant treatment. This suggests that RB1 mutations are not a common mechanism of resistance to palbociclib. Nonetheless, other alterations could lead to impaired RB function and signaling, such as epigenetic modifications and mutations in cis-regulatory regions resulting in decreased RB1 expression, or mutations in other genes of the RB pathway that were not included in the targeted sequencing panel. Of note, day 1 ctDNA analysis detected a high frequency of RB1 loss with no significant increase at the EOT. This high prevalence of RB1 loss may reflect an acquired mechanism of resistance to first-line endocrine treatment in metastatic disease, given that RB1 loss was shown to be a rare event in primary luminal-type breast cancers (3). These results could also imply that RB1 loss may be a mechanism of primary resistance to palbociclib after first-line endocrine treatment. These findings are intriguing, but will need to be further assessed in studies that include larger sample sets with matching germline DNA.

The sequencing analysis also demonstrated that PIK3CA mutations and the ESR1 Y537S mutation were acquired during treatment with either palbociclib and fulvestrant or fulvestrant alone. PIK3CA mutations are highly prevalent in primary and metastatic ER⁺ breast cancers. The BELLE-2 and BELLE-3 clinical trials showed that patients with metastatic breast cancer who had mutant PIK3CA ctDNA had improved PFS with PI3K inhibitors. Further, in these clinical trials, patients with PIK3CA-mutant ctDNA had numerically decreased PFS with fulvestrant treatment, reviewed in ref. 10. In this issue's study by O'Leary and colleagues, the acquisition of PIK3CA mutations during fulvestrant treatment aligns with the notion that PIK3CA mutations are a mechanism of endocrine resistance in metastatic disease.

The recurrent ESR1 mutations, which encompass a number of missense mutations mainly clustered in a hotspot in the ligand-binding domain, are the most common genomic mechanism of acquired resistance to endocrine treatments including fulvestrant (9). The observation by O'Leary and colleagues that the Y537S mutation was the only ESR1 mutation selected through treatment with fulvestrant alone or in combination with palbociclib suggests that this specific ESR1 mutation is a driver of resistance to fulvestrant. This result is consistent with emerging data showing that the different ESR1-mutant alleles are distinct and that the Y537S mutation displays the most robust phenotypes (11). Of interest, RB1 mutations were selected only in tumors with wild-type ESR1, in agreement with the concept that ER, CCND1/CDK4/6, and RB function in a linear pathway. Taken together, these results suggest that genetic mechanisms of endocrine resistance are important drivers of resistance to the combination of palbociclib with fulvestrant. These results underscore the need to better target the ER mutations and the PI3K pathway in combination with CDK4/6 inhibitors. This may become possible with the development of a new generation of oral selective estrogen receptor downregulators that effectively target the ESR1 Y537S mutation and newer PI3K inhibitors with better selectivity and toxicity profiles. However, whether more effective mutant ER inhibition will result in aberrant RB function and other downstream resistance mechanisms is unknown and may present yet another clinical challenge.

In summary, by studying serial cfDNA samples, O'Leary and colleagues identified three acquired drivers of resistance to fulvestrant and palbociclib, including RB1-inactivating mutations, PIK3CA, and ESR1-activating mutations. The latter two are associated with resistance to fulvestrant. Overall, this study presents a proof of concept for the value of ctDNA in monitoring the evolution of metastatic disease and provides important insights into the mechanisms of resistance to fulvestrant and palbociclib in the clinic. However, this study also portrays a number of challenges, including (i) the need for larger cohorts, particularly taking into consideration that not all specimens provide sufficient material for sophisticated analyses; and (ii) the current limited knowledge of the relevant gene networks involved in endocrine and palbociclib resistance. Advances in these challenges will help to guide the design of future gene or other molecular panels as new biomarker tools to be incorporated into different clinical trials tailored to the patient population and the study drugs. Although studies in the metastatic setting allow us to define biology, the real progress in patient outcomes will occur with the treatment of early-stage disease in the adjuvant setting when the complexity of clonal evolution has not yet emerged.

R. Schiff reports receiving commercial research grants from AstraZeneca and Gilead (grants to institute) and is a consultant/advisory board member for Eli Lilly and MacroGenics. R. Jeselsohn reports receiving research funding from Pfizer.

This article was supported by The Breast Cancer Research Foundation, BCRF-17-143 (R. Schiff) and NCI Grant K08 CA191058-02 (R. Jeselsohn).