Diagnostic Accuracy of New and Old Cognitive Screening Tools for HIV–Associated Neurocognitive Disorders

M Trunfio; D Vai; C Montrucchio; C Alcantarini; A Livelli; MC Tettoni; G Orofino; S Audagnotto; D Imperiale; S Bonora; G Di Perri; A Calcagno

Disclosures

HIV Medicine. 2018;19(7):455-464. 

In This Article

Results

Population

A total of 650 patients were enrolled in the study. A flow chart for the participants is depicted in Figure 1. The demographic and viroimmunological parameters of the whole population are shown in Table 1. It was found that 70 of 291 (24.0%), 69 of 441 (15.6%), 168 of 624 (26.9%) and 251 of 607 patients (41.3%) presented an altered FAB, CDT, 3QT and IHDS score, respectively. In total, 21.5% (140 of 650) presented with HAND at the further evaluation. On the basis of IHDS scores and the presence of neurological complaints, 300 patients were referred for the neurocognitive evaluation, and 281 screened patients underwent the evaluation (loss to retention of 6.3%) within an average of 1 month; their clinical and demographic characteristics as well as screening results are shown in Table 1. In this subgroup, 27 of 120 (22.5%), 34 of 119 (28.6%), 111 of 267 (41.6%) and 150 of 257 patients (58.4%) presented an altered FAB, CDT, 3QT and IHDS score, respectively. After the evaluation, 141 patients (50.2%) had normal results, and 102 (36.3%) were diagnosed with ANI, 32 (11.4%) with MND and six (2.1%) with HAD. The HAND prevalence was 49.8%. Compared with the whole population, those who underwent the full neurocognitive evaluation presented with higher frequencies of cognitive complaints and had a lower median IHDS score, as expected according to our selection criteria, a slightly lower median CD4 count nadir and education level, and a lower frequency of undetectable plasma HIV RNA.

Figure 1.

Flow–chart of the study from the enrollment to the neurocognitive evaluation. FAB, frontal assessment battery; 3QT, three questions test; IHDS, international HIV dementia scale; CDT, clock–drawing test; NC, neurocognitive.

Screening for HAND

The findings for the cross–tabulation of the index test results with the reference standard results, including the number and percentage of true positives/negatives and false positives/negatives, are shown in Table 2. The accuracy measures of the four tests are shown in Table 3. FAB proved to have the highest SP, PPV, CCR and Y–I, to the detriment of its SE. In comparison, IHDS presented the highest SE and NPV, but lower SP and slightly lower PPV, CCR and Y–I. CDT and 3QT performed poorly, with the lowest CCR and Y–I. The overall proportions of correctly classified patients were as follows: 68.3%, 65.4%, 53.8% and 45.7% for FAB, IHDS, CDT and 3QT, respectively. The highest proportions of false positives were found for 3QT and IHDS (22.8% and 22.2%, respectively), while the highest proportions of false negatives were observed for CDT and 3QT (31.9% and 31.5%, respectively), followed by FAB (29.2%). According to the LR+, only a positive FAB or IHDS result would increase by 8.3 and 1.7 times the pre–test odds of having HAND, respectively, while administering 3QT and CDT would not add anything more to the pre–test probability. As confirmation, having an altered IHDS or FAB score was found to be a significant risk marker for being subsequently diagnosed with HAND [P < 0.0001 for both; IHDS: OR 3.82, 95% confidence interval (CI) 2.3–6.5; FAB: OR 13.26, 95% CI 3.9–43.4]. CDT, IHDS and 3QT performance measures did not change significantly when evaluated in the 163 patients with plasma viral load < 20 HIV–1 RNA copies/mL (data not shown), while, interestingly, FAB SE (48.4%; 95% CI 32.0–65.2%), NPV (67.4%; 95% CI 53.4–78.8%), CCR (71.6%) and Y–I (0.40) increased moderately, whereas SP (91.7%; 95% CI 78.2–97.1%) and PPV (83.3%; 95% CI 60.8–94.2%) decreased slightly. In the ROC analysis, FAB and IHDS yielded notable AUROC values: 0.81 (95% CI 0.73–0.87; P < 0.001) and 0.73 (95% CI 0.67–0.78; P < 0.001), respectively, as shown in Figure 2. CDT presented an AUROC of 0.56 (95% CI 0.47–0.65; P = 0.23). We did not perform ROC analysis for 3QT in view of the nature of the test itself. Alternative cut–off points were explored, as shown in Table 4. Finally, screening accuracy was not significantly improved by combining two or more of the evaluated tools (data not shown).

Figure 2.

Receiver operating characteristic (ROC) curves: normal versus HIV–Associated Neurocognitive Disorders. FAB AUROC was 0.81 (95CI: 0.73–0.87; p <0.001); IHDS AUROC was 0.73 (95CI: 0.67–0.78; p<0.001). CDT AUROC was 0.56 (95CI: 0.47–0.65; p 0.23). Due to the nature of 3QT no ROC analysis was performed.

Screening for symptomatic HAND

The findings for the cross–tabulation of the index test results with the reference standard results, including the percentage of true positives/negatives and false positives/negatives, are shown in Table 2. The accuracy measures of the four tests in screening only MND/HAD are shown in Table 3. Once again, FAB had the highest SP, a better SE and an improved CCR and Y–I. Changes were observed in the same directions for 3QT, although its performance did not improve significantly. Surprisingly, IHDS showed a moderate decrease in its SE and a significant decrease in its SP, with an overall worse performance. Regarding CDT, although the SP was still acceptable, the SE was reduced further, resulting in an overall poor performance. The highest proportions of false positives were produced by IHDS and 3QT (48.6% and 35.9%, respectively), while the highest proportions of false negatives were observed for CDT and 3QT (9.2% and 7.9%, respectively). Consistent with this, only having an altered FAB score was found to be associated with a higher risk of MND/HAD diagnosis (P < 0.0001; OR 8.4; 95% CI 3.0–23.2).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....