Abstract 1732: Accuracy of extrapolation of circulating tumor cell count from small blood volumes: statistical estimation using the AccuCyte - CyteFinder system

Background: Baseline and post-treatment counts of circulating tumor cells (CTCs) are prognostic of patient outcome in cancer. The FDA-cleared CellSearch® system has defined CTC count as the number of cells per 7.5 mL of blood, with poor prognosis being 5 or more for breast and prostate cancers, and 3 or more for colon cancer. There are little reported data on whether it is reasonable to extrapolate CTC counts from smaller blood volumes, particularly when CTC count is low, given sampling variability. The AccuCyte - CyteFinder system (RareCyte) collects 7.5 mL of blood for analysis, isolates the buffy coat and smears it onto 8 slides, which are then stained with markers for CTCs that are identified by fluorescence microscopy. We investigated the statistical accuracy of extrapolation of total CTC count from fewer than 8 slides. Methods: 172 blood samples from various cancers were processed to microscope slides, stained, and analyzed by CyteFinder. Counts were made on a per slide basis, with total “true” CTC count being the sum of all 8 slides. Samples with no CTCs were excluded. To extrapolate the total CTC count from k slides, all possible combination of k slides were used to estimate the total count of 8 slides. This was done for k = 1-7 for all 172 samples and extrapolated counts were rounded to the nearest integer. The proportion of times that extrapolated counts fell within a specified percentage of the true CTC count was calculated across the entire sample set, as well as for 6 sub-categories defined by true CTC count range. Results: The 6 CTC count categories were: 1-4 (N=39), 5-10 (N=21), 11-25 (N=21), 26-50 (N=14), 51-100 (N=24), more than 100 (N=53). Cumulative proportions within 5, 10, 25 and 50% of the true total CTC count were determined for the entire sample set as well as by category. In the entire sample set, the proportion of extrapolated counts that fell within 25% of the true count ranged from 0.448 (1 slide) to 0.723 (4 slides) to 0.951 (7 slides). Generally, the lower the true total count, the lower the fraction of extrapolated counts that fell within the specified percentage of the true count. For example, for true count 5-10, the fraction that fell within 25% of the true count ranged from 0.208 (1 slide) to 0.513 (4 slides) to 0.951 (7 slides). In contrast, for true count more than 100, the fraction that fell within 25% of the true count ranged from 0.751 (1 slide) to 0.968 (4 slides) to 1.000 (7 slides). Conclusions: In this sample set, nearly three quarters of cases could be estimated within 25% of the true CTC count by extrapolating from 4 AccuCyte slides. Estimating each slide to represent ~1 mL of processed whole blood, our analysis suggests that a ~4 mL sample will have this level of accuracy relative to a CTC count from 7.5 mL. At low true CTC counts the accuracy of extrapolation decreases, indicating that a larger volume of blood is required for this level of accuracy. Citation Format: Jeffrey L. Werbin, Paulina Varshavskaya, Arturo B. Ramirez, Jackie L. Stilwell, Daniel E. Sabath, Ping-Yu Liu, Eric Kaldjian. Accuracy of extrapolation of circulating tumor cell count from small blood volumes: statistical estimation using the AccuCyte - CyteFinder system [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1732. doi:10.1158/1538-7445.AM2017-1732