Abstract 6238: Profiling of the circulating cell-free DNA methylome for detection and subtyping of small cell lung cancers

• Published in: Cancer research (Chicago, Ill.) Ill.), 2022-06, Vol.82 (12_Supplement), p.6238-6238
• Main Authors: Rothwell, Dominic G., Chemi, Francesca, Pearce, Simon, Clipson, Alex, Hill, Steven, Conway, Alicia Marie, Richardson, Sophie, Murat, Katarzyna, Caeser, Rebecca, Egger, Jacklynn, Poirier, John T., Kerr, Alastair, Blackhall, Fiona, Rudin, Charles M., Dive, Caroline
• Format: Article
• Language: English
• ISSN: 1538-7445; 1538-7445
• DOI: 10.1158/1538-7445.AM2022-6238

Introduction: Small-cell lung cancer (SCLC) is a high-grade neuroendocrine carcinoma characterised by high proliferation rate and early, rapid metastatic spread. Although SCLC is treated as a homogenous disease, recent studies revealed morphologic and transcriptomic heterogeneity with several molecular subtypes described based on predominant transcription factor expression (ASCL1, NEUROD1, ATOH1, POU2F3, YAP1) (Rudin et al., 2019; Simpson et al., 2020) which in preclinical studies exhibit differing vulnerabilities raising the potential of stratified therapy. DNA methylation is also thought be an important regulator of SCLC biology (Gazdar et al., 2017) and epigenetically distinct subtypes derived from SCLC primary tumour samples reported (Poirier et al., 2015). Here, we developed a robust workflow for genome-wide DNA methylation profiling to examine the potential use of cfDNA methylation profiling for detection and subtyping of SCLC. Results: To evaluate SCLC genome-wide DNA methylation patterns we employed a bisulfite-free enrichment-based approach (T7-MBD-seq). We tested this approach on tissue samples from preclinical models and from normal lung (n=110) and on cfDNA samples from both patients with SCLC and from non-cancer controls (n=157). Methylation profiles from preclinical models (patient-derived xenografts (PDX) and CTC derived explant (CDX) models) were comparable to previously described methylation patterns from SCLC primary tumours and were recapitulated in patients’ cfDNA samples. A tumour/normal classifier, based on 4,061 genomic regions detected as being hypermethylated in SCLC preclinical models, correctly assigned 93% and 100% cfDNA samples from patients with limited and extensive stage SCLC respectively, with a statistically significant correlation of prediction scores with disease stage (P=0.0076). Finally, to determine whether cfDNA methylation profiling could subtype SCLC patients, we built a subtype classifier, based on methylation signatures derived from 59 established SCLC cell lines. We applied the classifier to cfDNA samples from 56 patients and 10/11 with known subtypes (identified from a donor matched CDX model) were correctly classified. Overall, 73% of cfDNA samples were classified as ASCL1, 13% were classified as NEUROD1 and 14% were classified as being double negative with the distribution of the subtypes correlating closely to previously published IHC data from SCLC tissue samples (Baine et al., 2020). Conclusions: Our data