Abstract 339P
Background
Biomarkers that can predict chemotherapy response are much needed to improve and tailor treatment strategies in early stage colon cancer (CC). The consensus molecular subtype (CMS) classification divides CC in four biologically distinct subtypes and holds great promise as a predictive biomarker. To realize implementation into clinical practice, robust classification methods are needed and additional data is warranted on the predictive value of CMS for therapy response.
Methods
Two CMS identification methods were evaluated. First, we analysed the association between four histopathologic markers (tumour infiltrating lymphocytes (TILs), amount of mucus, tumour stroma ratio (TSR) and tumour budding) and CMSs in 218 early stage CC patients with available RNA-based CMS labels. Second, we generated NanoString and RNA sequencing profiles of 28 paired fresh frozen (FF) and formalin-fixed paraffin-embedded (FFPE) CC samples, plus 168 FF samples from an additional CC cohort, in order to develop a NanoString classifier. The gold standard CMS classifier was used as reference.
Results
TILs and a mucinous phenotype (>50% mucus surface) were positively correlated with CMS1 (p < 0.001 and p = 0.008). The presence of mucus (≥ 10%) was strongly associated with CMS3, mucus was present in 64.1% of CMS3 tumours. Higher budding counts were seen in CMS4 (p = 0.045). Although a strong association between TSR and CMS4 was established, still 52.5% (32/61) of CMS4 tumours was scored as stroma-low, indicating that CMS4 tumours cannot be identified by stromal content. A high classification accuracy was established for the developed NanoString classifier (93% in FFPE samples and 95% in FF samples), highlighting its possible utility in clinical practice.
Conclusions
CMS identification by histopathologic markers proved to be infeasible. As a next step we developed and validated a robust NanoString classifier suitable for FF and FFPE samples. The validity and predictive value for chemotherapy response of this classifier is currently being tested in a retrospective CC cohort and a prospective clinical trial applying neoadjuvant chemotherapy in early stage CC (CONNECTION-II). Preliminary results are promising and demonstrate subtype-specific responses to chemotherapy.
Clinical trial identification
NL8177.
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Dutch Cancer Society, Alpe d’HuZes.
Disclosure
G. Vink: Financial Interests, Institutional, Research Grant: BMS, Merck, Servier, Personal Genome Diagnostics, Bayer, Sirtex, Pierre Fabre, Lilly. M. Koopman: Financial Interests, Institutional, Advisory Board, advisory board and speaker: Pierre Fabre; Financial Interests, Institutional, Advisory Board: MSD, Bayer; Financial Interests, Institutional, Advisory Board, Advisory Board, speaker: Servier; Financial Interests, Institutional, Invited Speaker: Merck, BMS, Servier; Financial Interests, Institutional, Research Grant: Servier, Roche, Bayer, Bristol Myers Squibb, Merck, Personal Genomics Diagnostics, Sirtex, Pierre Fabre; Financial Interests, Institutional, Funding: Pierre Fabre, Amgen, Nordic Farma, Novartis, Merck, Servier, BMS; Non-Financial Interests, Leadership Role, vice-chair of DCCG: Dutch Colorectal Cancer Group; Non-Financial Interests, Advisory Role, Member of KWF scientific board: KWF; Non-Financial Interests, Other, ESMO faculty member for the Gastro-Intestinal Tumours – colorectal cancer: ESMO; Non-Financial Interests, Advisory Role, expert member of committee “regie op registers dure geneesmiddelen” ZINNL: ZiNNL; Non-Financial Interests, Advisory Role, CRC expert on Kanker.nl platform for answering online CRC questions of CRC (non) patients: Patient representative organisation (Kanker.nl); Non-Financial Interests, Leadership Role, chair of RWD & DH working group: ESMO; Other, Other, PI of the Dutch Prospective Colorectal Cancer Cohort study: PLCRC project. All other authors have declared no conflicts of interest.