Fig 1.
A. LightCycler real-time PCR amplification curves in a serial dilution assay. B-D. PCR efficiencies were determined in triplicate in two/three independent experiments. PCRs were performed in triplicate on five/six samples representing a 10-fold serial dilution. All standard curves demonstrated a linear relationship between the logarithm of the copy number and the Cp value, allowing determining the PCR Efficiency using: Efficiency = -1+10exp(-1/slope).
Fig 2.
Venn diagram of number of analyzed samples.
Inhibited samples: Samples were considered inhibited if the Cp of the pathogen-specific control was >38 for both wells for Toxoplasma-PCR (N = 45), and if ΔCp ≥3 for Pneumocystis-PCR (N = 17). Cp(albumin)>threshold: For albumin-PCR, the inhibition threshold was the mean Cpalb + 3 for a given matrix (see Methods and Table 1).
Table 1.
Frequency of PCR inhibition as a function of sample matrix for Toxoplasma PCR and Pneumocystis PCR.
Fig 3.
Youden’s index, sensitivity and specificity of albumin-PCR cut-off values.
A and C. Black curves: Youden’s index = sensitivity + specificity– 1; B and D. Red curves: sensitivity; blue curves: specificity. For Toxoplasma qPCR (A), the maximal Youden’s index was very low. The maximum value was 0.19 and was reached for a Cpalb cut-off value of 36. At this albumin qPCR cut-off value sensitivity is 20% (p< 0.0001; Binomial test for 95% theoretical sensitivity) and specificity 98.8% (p = 1, Binomial test for 95% theoretical specificity). (B) Graphical analysis of sensitivity and specificity curves in function of cut-off values. Both curves intersect at a Cpalb value of 22. At this Cpalb cut-off value sensitivity is 33.3% and specificity 66.6% (p<0.0001 and p<0.0001; Binomial tests for 95% theoretical sensitivity and specificity). For Pneumocystis qPCR (C), the maximal Youden’s index was too low to allow determining any Cpalb cut-off value. (D) Graphical profiles of sensitivity and curves in function of cut-off values. Both curves intersect at a Cpalb value of 22. At this Cpalb cut-off value sensitivity is 47.1% and specificity 62% (p<0.0001 and p<0.0001; Binomial tests for 95% theoretical sensitivity and specificity).
Fig 4.
Correlation between leucocyte counts and Cpalb values in 1690 blood samples.
Semi-log scale graph where leucocyte count (x-axis) was plotted against Cpalb (orange and green dots corresponding to amplification control inhibited and not inhibited samples respectively). Linear model was used to determine trend line (Blue) with 95% confidence interval (Grey).