Fig 1.
The methodological approach taken to evaluate the circulating gene transcript assay in PSPs compared to standard liquid-based PCR. GEP-NET = gastroenteropancreatic NET. PSP = pre-spotted plate.
Table 1.
Details of NETest genes (n = 51).
Fig 2.
Concordance between CQ and normalized gene expression levels in Sample Set I (n = 44).
2A/B. CQ levels. Scatter plot based on CQ for each of the target genes (genes are individually colored) on spotted plates versus the standard approach (2A). The individual Pearson values ranged from 0.29–0.98. The Tukey box and whisker’s plot of the individual r-values (2B) identify the mean was 0.76 and median 0.82. 2C/D. Gene expression. Scatter plot based on normalized gene expression for each of the target genes (individually colored) on spotted plates versus the standard liquid approach (2C). The individual Pearson values ranged from 0.21–0.98. The Tukey box and whisker’s plot of the individual r-values identify the mean was 0.65 and median 0.65 (2D).
Fig 3.
Relationship between NETest scores in the standard approach and PSPs in Sample Set I (n = 44).
3A. The mean and median NETest scores were not significantly different (p = 0.674) between liquid (mean: 49.7; median: 47) and PSPs (mean: 48.5; median: 60). 3B. The Pearson correlation for the NETest score was 0.873 (95%CI: 0.78–0.93), p<0.0001. Blue = controls (n = 8), red = NETs (n = 36). 3C. Graph plot of the averaged log-transformed gene expression of matched blood-tissue pairs. Error bars indicate standard error of the mean. The dotted line represents the best linear fit line. Regression analysis identified R to range from 0.71–0.83 (n = 7, p<0.0001). ULN = upper limit of normal.
Table 2.
Concordance based on Predictive Modeling.
Fig 4.
NETest scores and CgA levels in an independent validation cohort (Sample Set II, n = 167).
4A. All 30 controls exhibited NETest scores ≤20. Twenty-eight (93%) of non-NET controls exhibited scores ≤20. Forty-eight (96%) of 50 NETs exhibited scores >20. 4B. The AUROC for differentiating NETs from non-NET controls was 0.93±0.02 (95%CI: 0.88–0.97), p<0.0001. 4C. Individual CgA measurements from the control, non-NET controls and NET samples using a standard CLIA ELISA approach. 4D. AUROC comparison between the NETest and CgA for differentiating NETs from non-NET controls. The difference in AUC was 0.38±0.06 (95%CI: 0.26–0.49). The z-statistic was 6.44, p<0.0001. ULN = upper limit of normal.
Fig 5.
Robustness of PSP to attain a NETest result in Sample Set III (n = 18).
5A. Individual scores from the 18 samples in each of the two laboratories. Scores were not significantly different. Horizontal line = mean value for each group. 5B. Individual NETest correlations between the two laboratories (matched samples: n = 18). The correlation (Pearson r) was 0.967 (p<0.0001).
Fig 6.
Reproducibility of the PSP assays.
6A. Inter-assay reproducibility in five different samples. Each sample was repeated 5–10 times. The scores for each sample exhibited CV of 0–10%. The average CV was 4.2%. Bars are Mean+SD. 6B. Intra-assay reproducibility. Thirty samples with NETest scores spanning the range of reported results were repeated (Result 1 vs. Result 2). The averaged CV for the 30 samples was 1.26%. Line = Mean; CV = co-efficient of variation.
Fig 7.
Robustness of PSP compared to CLIA-laboratory liquid sample results (Sample Set IV, n = 48).
7A. Individual scores from the 48 consecutive samples using either standard CLIA approach (liquid) or on spotted plates (PSP). Scores were not significantly different. Horizontal line = mean value for each group. 7B. Individual NETest correlations between scores on standard CLIA vs. PSPs (n = 48). The correlation (Pearson r) was 0.94 (p<0.0001). 7C. Individual CgA measurements from the same 48 consecutive samples using a standard CLIA ELISA approach. Horizontal line = mean value for each group. ULN = upper limit of normal.
Table 3.
Concordance between clinical sample outputs—Standard CLIA vs. PSP.