Diagnostic Accuracy of pCLE and NBI for Colorectal Polyps
Diagnostic Accuracy of pCLE and NBI for Colorectal Polyps
Table 2 lists estimates of sensitivity and specificity of pCLE and NBI of small colorectal lesions compared with gold standard traditional histopathology. There was an evidence of higher sensitivity for pCLE vs. NBI (86 vs. 64%, P=0.008) but lower pCLE specificity compared with NBI (78 vs. 92% P=0.027). Overall accuracy did not differ noticeably between pCLE and NBI (82% vs. 79%, P=0.59). Positive and negative likelihood ratios were 3.88 and 0.18 for pCLE and 7.66 and 0.39 for NBI Table 2.
These findings were also observed in the subset of 103 polyps of size between 1 and 5 mm, where sensitivity of pCLE was 84% as compared with 60% with NBI (P=0.013), specificity for pCLE was lower as compared with that of NBI (76% vs. 95%, P=0.008), and accuracy was identical for the two methods at 80% (P=1.00). Positive and negative likelihood ratios were 3.50 and 0.21 for pCLE and 11.60 and 0.42 for NBI for this subset of polyps between 1 and 5 mm in size. The sensitivity of NBI was noted to increase in the smaller subset of 35 polyps of size 6–9 mm but it was still numerically, though not statistically, lower than the sensitivity of pCLE (92% vs. 77%, P=0.18). Positive and negative likelihood ratios were 6.46 and 0.09 for pCLE and 3.59 and 0.29 for NBI for this group of 35 polyps.
Among 130 polyps, 94 lesions (72.3%) were found to have high quality pCLE videos, and were rated 4 or 5 by the observer on scale of 1–5, where 1 being poor and 5 being the best and equal to histopathology ( Table 3 ). In this subset of polyps, sensitivity of pCLE was significantly higher as compared NBI (91% vs. 66%, P=0.008), while specificity was lower (74% vs. 92%, P=0.049), and overall accuracy did not differ (83% vs. 79%, P=0.42). The positive likelihood ratio was 3.58 for pCLE and 7.75 for NBI, whereas the negative likelihood ratio was 0.11 for pCLE and 0.37 for NBI.
In the subset of 65 polyps (50% of total 130 polyps) with high-quality pCLE videos, and in which the diagnosis of pCLE and NBI concurred, the accuracy, sensitivity, specificity, PPV, and NPV of pCLE and NBI were all very high (all ≥94%, Table 4 ). The positive likelihood ratio was 30.87 and the negative likelihood ratio was 0.07 for this subset of polyps.
Comparison of pCLE vs. NBI for all 130 Small Colorectal Polyps
Table 2 lists estimates of sensitivity and specificity of pCLE and NBI of small colorectal lesions compared with gold standard traditional histopathology. There was an evidence of higher sensitivity for pCLE vs. NBI (86 vs. 64%, P=0.008) but lower pCLE specificity compared with NBI (78 vs. 92% P=0.027). Overall accuracy did not differ noticeably between pCLE and NBI (82% vs. 79%, P=0.59). Positive and negative likelihood ratios were 3.88 and 0.18 for pCLE and 7.66 and 0.39 for NBI Table 2.
These findings were also observed in the subset of 103 polyps of size between 1 and 5 mm, where sensitivity of pCLE was 84% as compared with 60% with NBI (P=0.013), specificity for pCLE was lower as compared with that of NBI (76% vs. 95%, P=0.008), and accuracy was identical for the two methods at 80% (P=1.00). Positive and negative likelihood ratios were 3.50 and 0.21 for pCLE and 11.60 and 0.42 for NBI for this subset of polyps between 1 and 5 mm in size. The sensitivity of NBI was noted to increase in the smaller subset of 35 polyps of size 6–9 mm but it was still numerically, though not statistically, lower than the sensitivity of pCLE (92% vs. 77%, P=0.18). Positive and negative likelihood ratios were 6.46 and 0.09 for pCLE and 3.59 and 0.29 for NBI for this group of 35 polyps.
Comparison of pCLE vs. NBI in 94 Polyps With High Quality pCLE Videos
Among 130 polyps, 94 lesions (72.3%) were found to have high quality pCLE videos, and were rated 4 or 5 by the observer on scale of 1–5, where 1 being poor and 5 being the best and equal to histopathology ( Table 3 ). In this subset of polyps, sensitivity of pCLE was significantly higher as compared NBI (91% vs. 66%, P=0.008), while specificity was lower (74% vs. 92%, P=0.049), and overall accuracy did not differ (83% vs. 79%, P=0.42). The positive likelihood ratio was 3.58 for pCLE and 7.75 for NBI, whereas the negative likelihood ratio was 0.11 for pCLE and 0.37 for NBI.
Comparison of Combined pCLE/NBI With Histopathology
In the subset of 65 polyps (50% of total 130 polyps) with high-quality pCLE videos, and in which the diagnosis of pCLE and NBI concurred, the accuracy, sensitivity, specificity, PPV, and NPV of pCLE and NBI were all very high (all ≥94%, Table 4 ). The positive likelihood ratio was 30.87 and the negative likelihood ratio was 0.07 for this subset of polyps.
Source...