ctDNA dynamics: a novel indicator to track resistance in metastatic breast cancer treated with anti-HER2 therapy

Abstract

Background: Most studies utilizing circulating tumor DNA (ctDNA) to monitor disease interrogated only one or a few genes and failed to develop workable criteria to inform clinical practice. We evaluated the feasibility of detecting resistance to anti-HER2 therapy by serial gene-panel ctDNA sequencing.

Results: Primary therapeutic resistance was identified in 6 out of 14 patients with events of progressive disease. For this subset comparison of pre- and post-treatment ctDNA assay results revealed that HER2 amplification concurred with disease progression (4/6, 66.7%). Mutations in TP53 (3/6, 50.0%) and genes implicated in the PI3K/mTOR pathway (3/6, 50.0%) were also dominant markers of resistance. Together, resistance to HER2 blockade should be indicated during treatment if any of the following situations applies: 1) recurrence or persistence of HER2 amplification in the blood; 2) emergence or ≥20% increase in the fraction of mutations in any of these resistance-related genes including TP53/PIK3CA/MTOR/PTEN. Compared with CT scans, dynamic ctDNA profiling utilizing pre-defined criteria was sensitive in identifying drug resistance (sensitivity 85.7%, specificity 55.0%), with a concordance rate up to 82.1%. Besides, the ctDNA criteria had a discriminating role in the prognosis of HER2-positive metastatic breast cancer.

Methods: 52 plasma samples were prospectively collected from 18 patients with HER2-positive metastatic breast cancer who were treated with an oral anti-HER1/HER2 tyrosine kinase inhibitor (ClinicalTrials.gov NCT01937689). ctDNA was assayed by gene-panel target-capture next-generation sequencing.

Conclusions: Longitudinal gene-panel ctDNA sequencing could be exploited to determine resistance and guide the precise administration of anti-HER2 targeted therapy in the metastatic setting.

Keywords: anti-HER2 therapy; circulating tumor DNA; dynamics; metastatic breast cancer; progression.

Figure 1. Serial monitoring of genomic alterations in ctDNA

(panel A, patient No.3) A typical case illustrates the relationship between fluctuation patterns of HER2 copy number (right Y axis) and dynamics of tumor load (left Y axis). Notably, HER2 amplification in ctDNA was identified 8 weeks earlier than the clinical establishment of disease progression by CT. (panel B, patient No.2) The tumor load moderately decreased after C2 whereas HER2 copy number was elevated, which was followed by immediate disease progression after C4. (panel C, patient No.17; panel D, patient No.5; panel E, patient No.8) Notable increase in HER2 copy number and tumor burden was concurrently detected, regardless of HER2 status at baseline. (panel F, patient No.5) Dynamic ctDNA profiling revealed intra-tumor heterogeneity and clonal evolution, as evidenced by the diverging patterns of fluctuation in identified mutations. The left Y axis refers to the allele fractions of mutations in genes TSC2/MSH5/BRAC2 and the right Y axis to genes PIK3CA/FLT1/TP53.

Figure 2. Dynamics of somatic mutations in ctDNA

Fluctuations in allele fraction of somatic mutations (right Y axis) generally correlated with tumor burden reflected by imaging method (left Y axis), with increased allele fraction concurring with or even preluding disease progression (panel A., patient No.3; panel B., patient No.17).

Figure 3. Distribution of genomic patterns of resistance to anti-HER2 therapy

Relevant mutations involve genes TP53/PIK3CA/MTOR/PTEN, all of which have been identified to correlate with resistance to anti-HER2 therapy. *Patient No. 6 displayed neither HER2 amplification nor specific mutations but we captured a mutation in RSP14 prior to progression which might be associated with treatment failure.

Figure 4. Progression-free survival (PFS) of patients with events of progressive disease (N=14)

Based on sequencing data of C2 ctDNA, patients were evaluated as non-resistant (N=6) or resistant (N=8) using pre-defined ctDNA criteria. Non-resistance subset includes patient No.4, 6, 7, 15, 17, 18, and resistance subset includes patient No. 2, 3, 5, 8, 9, 11, 12, 14. Median PFS was 32.4 and 8.5 weeks respectively. P=0.0007 by log-rank test.