Oops, you're using an old version of your browser so some of the features on this page may not be displaying properly.

MINIMAL Requirements: Google Chrome 24+Mozilla Firefox 20+Internet Explorer 11Opera 15–18Apple Safari 7SeaMonkey 2.15-2.23

Poster Discussion session - Basic science

4745 - Functional Cell Profiling (FCP) of _100,000 CTCs from multiple cancer types identifies morphologically distinguishable CTC subtypes within and between cancer types


29 Sep 2019


Poster Discussion session - Basic science


Adam Jendrisak


Annals of Oncology (2019) 30 (suppl_5): v797-v815. 10.1093/annonc/mdz269


A. Jendrisak, R. Sutton, S. Orr, D. Lu, J. Schonhoft, Y. Wang

Author affiliations

  • Translational Research, Epic Sciences, 92121 - San Diego/US


Login to get immediate access to this content.

If you do not have an ESMO account, please create one for free.

Abstract 4745


Different cancers subtypes can often be effectively treated with similar Rx classes, i.e., with PARPi in the case of BRCA mutations or with anti-PD1/PDL1 inhibitors for patients with mismatch repair deficiencies. Using the Epic Sciences Functional Cell Profiling (FCP) platform utilizing high resolution digital pathology and machine learning, we analyzed ∼100,000 single CTCs from multiple indications to index metastatic cancers, with the longer-term hypothesis that sub-classification of CTCs may help define therapeutic decisions across cancer types.


92,300 CTCs underwent FCP analysis (single cell digital pathology features of cellular and sub-cellular morphometrics) were collected from cancer patients comprising Prostate (1641 pts, 70,747 CTCs), Breast (268 pts, 8,718 CTCs), NSCLC (110 pts, 1884 CTCs), SCLC (141 pts, 8,872 CTCs) and Bladder (65 pts, 2079 CTCs). An equal number of CTCs from each indication was sampled to create a balanced cohort for training. K-means clustering was applied on the training set and an optimized number of clusters were determined using the elbow approach. After defining the clusters on the training set, the cluster centers were extracted from k-means, and used to train a k-Nearest Neighbor (k-NN) classifier to predict the cluster assignment for the remaining CTCs designated as the test set.


The optimized number of clusters was 9. Breast cancer CTCs were more enriched with higher CK expressing CTCs (c1), while SCLC and some of mCRPC shared similar small cell like features (c5). Cluster subtypes and characteristics are listed.Table:


c1Highest CK8%24%1%5%9%
c2Highest N/C ratio13%15%7%35%21%
c3Large, Elongated1%3%12%1%3%
c4Low N/C, Low CK33%24%25%8%12%
c5Small, Higher N/C ratio, More Round11%12%11%37%33%
c6Low CK, Elongated8%4%8%2%4%
c7Highest Nucleoli Count0%0%17%1%4%
c8Large Nuclear Area, Highest Nuclear Entropy21%14%10%11%10%
c9Most Elongated Cytoplasm4%3%9%2%4%


Heterogeneous CTC phenotypic subtypes were observed across multiple indications. Each indication harbored subtype heterogeneity and shared clusters with other cancers. Patient cluster subtype analysis for prognosis and therapy benefit as well as linkage of CTC subtype genotypes to patient survival by cancer type is ongoing.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.


Epic Sciences.


A. Jendrisak: Full / Part-time employment: Epic Sciences. R. Sutton: Full / Part-time employment: Epic Sciences. S. Orr: Full / Part-time employment: Epic Sciences. D. Lu: Full / Part-time employment: Epic Sciences. J. Schonhoft: Full / Part-time employment: Epic Sciences. Y. Wang: Full / Part-time employment: Epic Sciences.

This site uses cookies. Some of these cookies are essential, while others help us improve your experience by providing insights into how the site is being used.

For more detailed information on the cookies we use, please check our Privacy Policy.

Customise settings
  • Necessary cookies enable core functionality. The website cannot function properly without these cookies, and you can only disable them by changing your browser preferences.