Size-based isolation and detection of renal carcinoma cells from whole blood

Abstract

Renal cell carcinoma (RCC) is a tumour type with an indolent growth pattern and rather vague symptoms. The present study developed a platform for liquid biopsy of RCC based upon the isolation of circulating tumour cells (CTCs). Founded on the observation that RCC tumour cells are considerably larger than leucocytes, the present study employed a microfluidics-based system for isolation of RCC CTCs from whole blood. Using this system, it was revealed that 66% of spiked-in RCC tumour cells could be retrieved using this approach. Furthermore, it was demonstrated that these cells could be molecularly detected with digital PCR using RCC-specific genes down to one tumour cell, whilst avoiding detection in samples lacking tumour cells. Finally, subtype specific transcripts were identified to distinguish the different subtypes of RCC, which were then validated in patient tumours. The present study established a novel workflow for the isolation of RCC CTCs from whole blood, with the potential to detect these cells irrespective of subtype.

Keywords: circulating tumour cells; kidney cancer; liquid biopsy; molecular diagnostics.

Figure 1

Cell diameters of primary RCC, cultured RCC and non-RCC cells. AO, Acridine Orange; ccRCC, clear cell renal cell carcinoma; p1RCC, papillary type 1 RCC; RBC, red blood cell lysis.

Figure 2

ClearCell FX recovery efficiency and immunofluorescence-based detection of RCC cells. (A) SNU-349 and primary RCC cell line spike-in and recover efficiencies (%). Horizontal lines represent average recovery efficiency and error bars are standard error of mean (SEM). Mann-Whitney U Test, ^*P<0.05 (two-tailed). (B) Immunofluorescence-based detection of spiked SNU-349 cells enriched from whole-blood using the ClearCell FX system. The output material from 7.5 ml of whole blood spiked with SNU349 cells were immobilized by cytospin centrifugation and labelled with antibodies against CD45 and CA9 for detection of leukocytes and ccRCC cells, respectively. Fluorophore signal is merged. Purple, CA9; Green, CD45; Blue, DAPI. RCC, renal cell carcinoma.

Figure 3

Specificity, reproducibility and detection of ccRCC cells with SLC6A3 and CA9. (A) Specificity of SLC6A3 and CA9 expression in RCC tumours within TCGA database. (n=103 ccRCC.e1, 255 ccRCC e.2, 137 ccRCC e.3, 77 chRCC, 134 p1RCCa, 71 p1RCCb, 52 p2RCC). ^****P<0.0001. (B) Digital PCR assay efficiency of SLC6A3 and CA9 over increasing pre-amplified cDNA inputs with line of best fit. (C) Digital PCR based copies/µl measurement of SNU-349 spiked blood samples after ClearCell FX enrichment and downstream processing. Copies/µl also shown for 4 enriched blood samples without spiked-in SNU-349 cells. ccRCC, clear cell renal cell carcinoma; chRCC, chromophobe RCC; p1RCC, papillary type 1 RCC; p2RCC, papillary type 2 RCC; cDNA, complementary DNA.

Figure 4

Relative expression of non-ccRCC markers derived from Limma analysis of TCGA expression data. Relevant subtypes are shown according to Chen et al (26) taxonomy. ccRCC cohorts and p1RCC cohorts grouped for Dunn's comparison test (n=103 ccRCC.e1, 255 ccRCC e.2, 136 ccRCC e.3, 77 chRCC, 134 p1RCCa, 71 p1RCCb, 52 p2RCC). ^***P<0.001, ^****P<0.0001. ccRCC, clear cell renal cell carcinoma; chRCC, chromophobe RCC; p1RCC, papillary type 1 RCC; p2RCC, papillary type 2 RCC.

Figure 5

qPCR based mRNA expression relative to beta-actin for all selected markers in patient primary RCC tumour tissue. Horizontal lines represent average mRNA expression and error bars represent SEM (n=4 for each subtype). Data were analysed using Kruskal-Wallis with post-hoc Dunn's test. ^*P<0.05. Comparisons not reaching significance not indicated. ccRCC, clear cell renal cell carcinoma; chRCC, chromophobe RCC; p1RCC, papillary type 1 RCC; p2RCC, papillary type 2 RCC.