Sampling circulating tumor cells for clinical benefits: how frequent?

Abstract

Circulating tumor cells (CTCs) are cells shed from tumors or metastatic sites and are a potential biomarker for cancer diagnosis, management, and prognostication. The majority of current studies use single or infrequent CTC sampling points. This strategy assumes that changes in CTC number, as well as phenotypic and molecular characteristics, are gradual with time. In reality, little is known today about the actual kinetics of CTC dissemination and phenotypic and molecular changes in the blood of cancer patients. Herein, we show, using clinical case studies and hypothetical simulation models, how sub-optimal CTC sampling may result in misleading observations with clinical consequences, by missing out on significant CTC spikes that occur in between sampling times. Initial studies using highly frequent CTC sampling are necessary to understand the dynamics of CTC dissemination and phenotypic and molecular changes in the blood of cancer patients. Such an improved understanding will enable an optimal, study-specific sampling frequency to be assigned to individual research studies and clinical trials and better inform practical clinical decisions on cancer management strategies for patient benefits.

Fig. 1

CTC response patterns in cancer patients undergoing therapy. a Hypothetical CTC count patterns corresponding to patient response to therapy. The light green circles represent frequent sampling times, and the dark green circles represent infrequent sampling times. b CTC count pattern of a breast cancer patient from response group 3 of Pachmann et al. [18]. CTC count patterns are simulated when the frequency of sampling is halved from six to three sampling points. The sampling points chosen for each simulation are numbered

Fig. 2

Case studies of CTC count patterns in cancer patients. a CTC count pattern of a metastatic breast cancer patient from Marsland and Schuur [19]. CTC count patterns are simulated when the frequency of sampling is halved. Alternate time points were used for each simulation. PD indicates progressive disease and PR indicates partial response. b Longitudinal monitoring of CTC count and PSA level in a prostate cancer patient. CTCs were measured using the CellSearch assay. CTC count patterns are simulated when the frequency of sampling is halved. Alternate time points were used for each simulation. PD indicates progressive disease. c Longitudinal monitoring of CTC count and PSA level in a metastatic prostate cancer patient. CTCs were imaged at ×20 magnification and identified as Hoechst (blue) positive and CD45 (green) negative, and white blood cells (WBCs) as both Hoechst and CD45 positive. Some CTCs were also EpCAM/cytokeratin (red) positive, while WBCs were EpCAM/cytokeratin negative. The foremost graph shows concentration of PSA with time. The second graph behind shows the CTC count over time. The third to fifth graphs behind are representatives of simulated models for monthly CTC counting. The green circles represent sampling times. The proportion of simulations represented by each model is indicated as percentage. A [¹⁸F]NaF-PET bone scan in July 2012 vs. April 2012 showed progression of osseous disease including new lesions (as indicated by blue arrows)

Fig. 3

Hypothetical models of different sampling times for molecular assays of CTCs. a CTC count pattern and resistance mutation (Mut X) detection in a hypothetical cancer patient on targeted therapy. Mut X indicates presence of the mutation at low allele frequency. CTC count patterns and resistance mutation detection are simulated when the frequency of sampling is increased from once in 2 months (red squares) to once in two weeks (green circles). The bars below show the impact of earlier resistance mutation detection on choice of therapy. The red bars indicate the therapeutic strategy with infrequent CTC sampling while green bars indicate the therapeutic strategy with frequent CTC sampling. Time saved will vary, depending on when Mut X is detected at high allele frequency between week 6 and 15. b CTC count pattern and epithelial-mesenchymal phenotypes in a hypothetical patient with CTCs oscillating between the epithelial and mesenchymal phenotypes. Frequent sampling time is indicated in red, while infrequent sampling times are simulated in blue (infrequent sampling 1) and green (infrequent sampling 2). Infrequent sampling 1 and 2 differ in their start time for first sampling. The proportion of mesenchymal phenotype is indicated in the bars below in blue while that of epithelial phenotype is indicated in white