Urinary transcript quantitation of CK20 and IGF2 for the non-invasive bladder cancer detection

Abstract

Purpose: Cytokeratin 20 (CK20) and insulin-like growth factor 2 (IGF2) were previously proposed to be elevated in clinical samples from patients with bladder cancer (BCa). A two cohort design validation study was used to assess the relevance for BCa detection by transcript quantitation of both markers in urine samples. Their diagnostic value was assessed in comparison with voided urine cytology (VUC).

Methods: RNA isolation was carried out using cellular sediments of urine samples from 196/103 histologically positive BCa patients, as well as 97/50 control subjects for the test (TC) and validation cohort (VC), respectively. Urinary transcript levels of CK20 and IGF2 were determined by qPCR.

Results: Relative transcript levels were significantly elevated 3.4/11-fold for CK20 and 188/64-fold for IGF2 (p < 0.001) in urine sediments of BCa patients compared to controls in the TC and VC, respectively. In a combined analysis, the resulting sensitivity (SN) (SN_TC: 77.9; SN_VC: 90.3%) and specificity (SP) (SP_TC: 88.0; SP_VC: 84.0%) were similar to that of VUC. The sensitivity of VUC in combination with CK20 and IGF2 was considerably increased (SN_TC: 94.6; SN_VC: 93.2%) while specificity was reduced (SP_TC: 72.0; SP_VC: 82.0%) compared to VUC alone in the test and validation cohort.

Conclusions: Transcript levels of IGF2 and CK20 enabled the detection of BCa with a diagnostic performance similar to VUC. Combined analysis of voided urine cytology together with altered transcript levels of CK20 and IGF2 enhanced sensitivity, but did not improve overall test performance.

Keywords: Cytokeratin 20; Insulin-like growth factor II; KRT20; Tumor marker; Urine; Urothelial carcinoma.

Fig. 1

Overview of patients and controls enrolled in the test and validation cohort. In the TC a total of 196 patients and 97 controls were enrolled. The results of the marker analyses of patients with new-onset disease (BCa primary) and a history of BCa (BCa recurrent) as well as control subjects with (NED BCa history) and without a history of BCa (control no tumor) in the TC were significantly different. As a result, individuals in the TC with a history of BCa (45 patients, 27 controls) were excluded from further analyses. In the VC 103 patients without a history of BCa and 50 controls were enrolled

Fig. 2

Relative transcript levels of CK20 (a) and IGF2 (b) in urine from control subjects and BCa patients with and without BCa history in the TC. Patients without histopathological tumor evidence and without a known BCa history, as well as healthy subjects (control no tumor) are analyzed in comparison to histopathologically tumor-free patients with a known history of BCa (NED BCa history). BCa patients with histopathological tumor evidence are subdivided according to primary (BCa primary) and recurrent BCa (BCa recurrent) diagnosis. Transcript levels were normalized to RPLP0. p values (Mann–Whitney U) are given above brackets. p values <0.05 were considered significant. Individual sample size (n) is given for each group

Fig. 3

Urinary transcript levels of CK20 and IGF2 in control subjects and BCa patients. Healthy subjects and patients, who underwent TUR-B without evidence of BCa, were allocated to the control groups in the TC (a, b). Patients with concomitant chronic cystitis (no tumor cystitis) were analyzed in comparison to controls without cystitis (no tumor) for CK20 (a) and IGF2 (b). Equally, TUR-B patients without evidence of BCa (no tumor) were compared to urolithiasis controls (no tumor urolithiasis) for CK20 (c) and IGF2 (d) in the VC. Furthermore, statistical differences between the complete control group and the BCa patients were calculated. Transcript levels of both markers were normalized to RPLP0. p values (Mann–Whitney U) are given above brackets. p values <0.05 were considered significant. Individual sample size (n) is given for each group

Fig. 4

Receiver operating characteristic curve analyses of urinary CK20 and IGF2 for the diagnostic discrimination between primary diagnosed BCa patients and controls in the TC (a) and VC (b). AUC area under the curve, C number of controls, BCa number of patients

Fig. 5

Correlation of relative CK20 (a, c) and IGF2 (b, d) transcript levels and tumor stage in the TC (a, b) and VC (c, d). Following the histopathological examination of the resected tumor tissue, BCa patients were classified on the basis of their tumor stage as pTa, pT1, pTis and ≥pT2. Patients with pTis only, as well as patients up to tumor stage pT1 with concomitant pTis were included in a separate pTis group. Transcript levels of CK20 and IGF2 were normalized to RPLP0. p values (Mann–Whitney U) for difference between controls and tumor stages are given above brackets. p values <0.05 were considered significant. Individual sample size (n) is given for each group

Fig. 6

Associations of relative transcript levels of CK20 (a, c) and IGF2 (b, d) and histopathological grading (WHO 1973) in urine from BCa patients in the TC (a, b) and VC (c, d). BCa patients were classified on the basis of the results of the histopathological tumor grading as G1, G2, and G3. Transcript levels of CK20 and IGF2 were normalized to RPLP0. p values (Mann–Whitney U) for difference between controls and WHO 1973 tumor grades are given above brackets. p values <0.05 were considered significant. Individual sample size (n) is given for each group

Fig. 7

Correlation of histopathological grading system (WHO 2004) and relative transcript levels in urine from BCa patients of CK20 (a, c) and IGF2 (b, d) for the TC (a, b) and VC (c, d). BCa patients were classified on the basis of the results of the histopathological tumor grading as low-grade and high-grade tumors. Transcript levels of CK20 and IGF2 were normalized to RPLP0. p values (Mann–Whitney U) for difference between controls and tumor grades are given above brackets. p values <0.05 were considered significant. Individual sample size (n) is given for each group