AR alterations inform circulating tumor DNA detection in metastatic castration resistant prostate cancer patients

Abstract

Circulating tumor DNA (ctDNA) in plasma cell free DNA (cfDNA) of cancer patients is associated with poor prognosis, but is challenging to detect from low plasma volumes. In metastatic castration-resistant prostate cancer (mCRPC), ctDNA assays are needed to prognosticate outcomes of patients treated with androgen receptor (AR) inhibitors. We develop a custom targeted cfDNA sequencing assay, named AR-ctDETECT, to detect ctDNA in limiting plasma cfDNA available from mCRPC patients in the Alliance A031201 randomized phase 3 trial of enzalutamide with or without abiraterone. Of 776 patients, 59% are ctDNA-positive, with 26% having high ctDNA aneuploidy and 33% having low ctDNA aneuploidy but displaying AR gain or structural rearrangement, MYC/MYCN gain, or a pathogenic mutation. ctDNA-positive patients have significantly worse median overall survival than ctDNA-negative patients (29.0 months vs. 47.4 months, respectively). Here, we show that mCRPC patients identified as ctDNA-positive using the AR-ctDETECT assay have poor survival despite treatment with potent AR inhibitors in a phase 3 trial.

Trial registration: ClinicalTrials.gov NCT01949337.

Fig. 1. Cell free DNA (cfDNA) yield and circulating tumor DNA (ctDNA) fraction in A031201 plasma specimens.

a CONSORT flow diagram of patients enrolled in A031201 and plasma cfDNA specimens analyzed by targeted DNA-seq. ‘n’ refers to the number of patients. b AR-ctDETECT targeted DNA-seq assay design. c Scatterplot of ctDNA aneuploidy fraction estimated by ichorCNA using off-target DNA-seq reads vs. cfDNA yield from 776 plasma specimens analyzed by targeted DNA-seq. Loess trendline and 95% confidence intervals are shown. Violin plots illustrate data density from min to max. Boxes in violin plots illustrate median and interquartile range. Whiskers illustrate 1.5X interquartile range. d ctDNA aneuploidy fraction estimate as in (c) using cfDNA isolated from pooled plasma from healthy donors. Violin plots illustrate data density from min to max. Boxes in violin plots illustrate median and interquartile range. Whiskers illustrate 1.5X interquartile range. An assay cutoff of mean + 3 standard deviations (0.142) distinguishes ctDNA aneuploidy-high specimens.

Fig. 2. Comprehensive profiling of AR genomic alterations in A031201 cfDNA.

a ctDNA aneuploidy fraction estimated by ichorCNA in targeted DNA-seq data from n = 776 cfDNA specimens. Samples below the ctDNA aneuploidy-high threshold of 0.142 are in gray/pink. b Relative AR gene copy number ratio in cfDNA specimens ordered left to right as in (a). A log2 ratio >= 0.3 is considered a gain. c Relative AR upstream enhancer copy number ratio in cfDNA specimens ordered left to right as in (a). A log2 ratio >= 0.3 is considered a gain. d Number of AR gene structural rearrangements (AR-GSRs) in cfDNA specimens ordered left to right as in (a). AR-GSRs predicted to encode AR variant proteins with a truncated ligand binding domain (LBD), and AR-GSRs predicted to be associated with AR amplification on extrachromosomal DNA (ecDNA), are shown. e AR single nucleotide variants (SNVs) detected in cfDNA specimens as in (a). f Scatterplot of relative AR gene copy number ratio and number of AR-GSRs. Dots are colored blue based on copy number gain of the AR gene body. Dark blue denotes samples with 2 or more AR-GSRs and AR gene body copy gain, which is a signature of AR ecDNA (shown also in (d)). ‘n’ refers to the number of patient samples. Forest plots illustrating hazard ratio (squares) and 95% confidence intervals (horizontal lines) for (g) radiographic progression (rPFS) and (h) overall survival (OS) in patients demonstrating indicated cfDNA features. Multivariable analysis is adjusted for ctDNA aneuploidy fraction. P-values are from the Wald test from the Cox’s proportional hazards model, adjusted for multiplicity using the Benjamini-Hochberg method (false discovery rate). A FDR < 0.05 is considered statistically significant. ‘n’ refers to the number of patients. i UpSet plot showing co-occurrence of indicated AR genomic alterations. j Scatterplot of relative AR gene copy number ratio and AR upstream enhancer copy number ratio. Dots are colored based on copy number gain of the AR gene body only (brown), AR upstream enhancer only (gold), or both the AR gene body and AR upstream enhancer (red). ‘n’ refers to the number of patient samples.

Fig. 3. MYC and MYCN copy number gains in A031201 cfDNA.

a Relative MYC and MYCN gene copy number ratios in cfDNA specimens based on ctDNA aneuploidy fraction estimated by ichorCNA (samples below the ctDNA aneupoidy-high threshold of 0.142 are in gray/pink). A log2 ratio >= 0.3 is considered a gain. b UpSet plot showing co-occurrence of MYC and MYCN copy number gains in cfDNA samples from (a). c UpSet plot showing co-occurrence of MYC and MYCN copy number gains in mCRPC biopsy specimens from 444 patients analyzed in the AACR-PCF Stand Up To Cancer East Coast study. d UpSet plot showing co-occurrence of MYC and MYCN copy number gains in mCRPC biopsy specimens from 101 patients analyzed in the AACR-PCF Stand Up To Cancer West Coast study.

Fig. 4. Prevalent genomic alterations in ctDNA-positive A031201 cfDNA.

UpSet plots showing relationships between ctDNA features and genomic alterations in (a) ctDNA aneuploidy-high Group 1 and (b) ctDNA aneuploidy-low Group 2. c cfDNA specimens stratified into 3 groups based on samples exceeding a ctDNA aneuploidy threshold (ctDNA aneuploidy-high Group 1), samples below the ctDNA aneuploidy threshold but harboring a likely somatic pathogenic mutation, an AR-GSR, and/or copy number gain of AR, MYC or MYCN (ctDNA aneuploidy-low Group 2), or samples below the ctDNA aneuploidy threshold and lacking a likely somatic pathogenic mutation, an AR-GSR, or copy number gain of AR, MYC or MYCN (ctDNA-negative Group 3). d Pie charts illustrating frequency of AR or MYC/MYCN alterations detected in ctDNA aneuploidy-high Group 1 or ctDNA aneuploidy-low Group 2. e TP53, PTEN, and RB1 single nucleotide variants (SNVs) detected in ctDNA-positive cfDNA specimens ordered left to right as in (c). Relative (f) TP53, (g) PTEN, or (h) RB1 gene copy number ratios in cfDNA specimens ordered left to right as in (c). A log2 ratio <= −0.3 in a ctDNA-positive sample is considered a loss. UpSet plots showing relationships between (i) TP53, (j) PTEN, or (k) RB1 SNVs and copy number losses in ctDNA-positive cfDNA samples. l Pie charts illustrating frequency of TP53, PTEN, or RB1 alterations detected in ctDNA aneuploidy-high Group 1 or ctDNA aneuploidy-low Group 2. m UpSet plot showing relationships between alterations in TP53, PTEN, and RB1 in ctDNA-positive cfDNA specimens.

Fig. 5. Pathogenic mutations detected in A031201 ctDNA.

Oncoprint of single nucleotide variants (SNVs) occurring in indicated genes across ctDNA specimens in ctDNA aneuploidy-high Group 1 and ctDNA aneuploidy-low Group 2. The frequency of SNVs within these groups is indicated at the right. Samples are ordered left to right by descending ctDNA aneuploidy fraction estimate.

Fig. 6. Survival of ctDNA-positive A031201 patients.

Kaplan-Meier plots of (a) radiographic progression-free survival and (b) overall survival in patients classified as ctDNA-positive or ctDNA-negative. Kaplan-Meier plots of (c) radiographic progression-free survival and (d) overall survival in patients classified as belonging to ctDNA aneuploidy-high Group 1, ctDNA aneuploidy-low Group 2, or ctDNA-negative Group 3.