CDC reports on two alternative HIV testing algorithms

However, there remains a need for better understanding of cost-effectiveness and feasibility and barriers to implementing this algorithm and to using viral load tests for diagnosis.

“Most importantly, we still need an FDA-approved quantitative viral load test that also has an indication for diagnosis of HIV-1 infection before an algorithm like this one could be implemented more widely.”

Silvina Masciotra, MS, research microbiologist with the CDC’s Division of HIV/AIDS Prevention, presented an alternative two-step algorithm that also used HIV-1 RNA as the second step. “HIV-2 infections are rare in the United States,” she said in reference to the HIV-1/2 differentiation test, which is step two in the CDC/APHL recommended three-step algorithm.

Despite the widespread adoption of the recommended algorithm, “fewer labs have implemented the use of the FDA-approved nucleic acid test for the detection of acute infection,” for workload reasons, Masciotra said. “Viral load is often used as an alternative third test, though it is not an intended use.”

The CDC took part in the clinical trial for the BioPlex 2200 HIV Ag-Ab assay, a multiplex flow immunoassay intended for the simultaneous detection of p24 and HIV-1 and HIV-2 antibodies in human serum or plasma. “We had the chance to evaluate seroconversion panels at the CDC,” Masciotra noted. Those evaluations found that the BioPlex assay performed similarly to any other FDA-approved laboratory-based antigen-antibody immunoassay in detecting HIV-1 early infection.

The Hologic Aptima HIV-1 Quant assay on the Panther system is FDA approved for viral load monitoring. “We know the same assay is approved outside of the United States with a dual claim” for HIV-1 diagnosis and monitoring, Masciotra said. “This means we can use the viral load assay for diagnosis by following the interpretation” provided in the package insert that is approved outside of the United States. “Everything that was detected, even if it was not quantified because it was lower than the limit of quantification, we called it HIV-1 RNA positive,” Masciotra said.

The Aptima Quant assay is a high-throughput, fully automated, test tube platform with random access and uses transcription-mediated amplification. The reported limit of detection is 12 copies/mL. Its linear range of quantification is 30–107 copies/mL. The CDC did an in-house evaluation of the Aptima Quant assay for diagnosis.

In a comparison study of 417 samples from U.S. seroconverters infected with HIV-1, the Aptima Quant detected virus in more samples including the seronegative phase than the Hologic Aptima HIV-1 RNA Qualitative Assay approved for diagnosis. In patients with established HIV-1 infection, both tests performed similarly. “In our validation, we concluded that both assays were equal.”

Why use HIV-1 RNA viral load as the second test in the algorithm? “Since we do a lot of evaluations in the laboratory, I had the data. I said, ‘What if you look at a two-test diagnostic algorithm using a screening test that gives you the ability to differentiate HIV-1 from HIV-2 antibodies and p24, followed by the HIV-1 nucleic acid test—in this case the Aptima Quant—that also has approval outside of the United States,’” and may be approved by the FDA someday.

The objective was to compare the performance of a two-test diagnostic algorithm, consisting of screening with an antigen-antibody HIV-1/2 differentiation immunoassay, followed by an HIV-1 nucleic acid test, to the currently recommended three-test algorithm.

The analysis of specificity for the BioPlex assay during the clinical trial was calculated using 596 HIV-negative samples and the CDC contributed some of the testing. For the Aptima Quant assay, the laboratory tested 478 HIV-1–negative samples. “Because contamination problems have been reported with viral load open platforms, we decided to do a carryover contamination experiment in an open platform,” Masciotra said.

In comparing the two algorithms, the CDC used 46 U.S. seroconverters (subtype B), with 255 longitudinal samples before antiretroviral therapy initiation and 73 samples after ART initiation. “All of those were BioPlex seroreactive, and they were all positive after the first HIV-1 RNA,” she said. The comparison also included 105 Cameroonian ART-naïve specimens with established HIV-1 infection; three had HIV-1 group O infection and 102 had HIV-1 group M, non-B subtype infection. The CDC conducted the evaluations as part of collaborations with Bio-Rad and Hologic, both of which provided kits.

The BioPlex assay had a specificity of 99.7 percent. Aptima Quant had a specificity of 99.8 percent and no carryover contamination was observed.

The three-step algorithm detected HIV-1 infection in 96.2 percent of 79 samples from seroconverters with early infection before ART initiation. There was good agreement between the BioPlex and Geenius assays regarding HIV-1 detection, and HIV-2 reactivity was not observed with either assay, Masciotra said. Results were confirmed with the Aptima qualitative assay.

The two-test alternative algorithm detected 98.7 percent with a viral load range of less than 1.47 to more than 7 log (cop/mL). “We had two samples that were detected but non-quantified at this early stage of infection, and two of the Aptima qualitative assay nonreactives were either less than 1.47 or 4.89 log (cop/mL).” When Masciotra performed the McNemar’s comparison analysis (p=0.4795), there were no significant differences. “Both algorithms performed similarly in early stages of infection before ART initiation,” she said.

In later stages of infection, with the three-test algorithm, 176 samples had HIV-1 positive results on the seroconversion panels with the Geenius assay. All Cameroonian samples with established HIV-1 infection were Geenius HIV-1 positive. “One sample was also HIV-2 untypable where it showed HIV-2 reactivity,” she said. “We did not see HIV-2 reactivity on the BioPlex assay.” Further testing found no evidence of HIV-2 infection.

Comparison of the two algorithms on samples with established HIV-1 infection before ART initiation revealed a detection rate of 99.3 percent with the two-step algorithm, Masciotra said. The viral load range was from less than 1.47 to greater than 7 log (cop/mL). Two samples were nonreactive with the Aptima Quant and the Aptima qualitative assays, and seven samples were detected and not-quantified later in infection.

“When we looked at the McNemar’s comparison analysis, there was similar performance between the two algorithms in established infection before ART initiation,” Masciotra said.

Turning to detection of HIV-1 infection immediately after ART initiation, the three-test algorithm detected HIV-1 in 100 percent of the 73 samples. Nine samples from seroconverters showed seroreversion. “People go through viral suppression over time. You will see seroreversion to even an HIV antibody negative. But those were Aptima qualitative assay positive,” Masciotra said.

The two-test algorithm showed a lower detection rate of 87.7 percent. The viral load range extended from target not detected to 6.9 log (cop/mL). Aptima Quant did not detect three of the nine samples that sero­reverted. Six of the 64 HIV-1 Geenius positive samples were target not detected on Aptima Quant.

“In this case, although the number of samples is limited, the McNemar’s analysis showed significant differences in the two-test algorithm,” Masciotra said. “It did not perform as well as the three-test algorithm.”

The study’s limitations were that there were not enough samples to repeat. The Aptima qualitative and quantitative assays were not performed in parallel for a set of ART-naïve seroconversion panels. The Geenius HIV 1/2 differentiation assay was performed using software version 1.1, which was prior to the update to address HIV-2 indeterminate results, though “there were no HIV-2 indeterminate results in our sample set,” she said. And there was a small number of samples from ART-treated persons.

Aptima Quant worked well for diagnosis and quantification as a second step in the proposed algorithm in different stages of HIV-1 infection, Masciotra said, though it is not FDA approved with a dual claim. The assay’s performance decreased after the IgG response is elicited and with suppressed viremia due to ART. “Maybe they can use an antibody test when there is an undetectable viral load result,” she said.

Confirmation with a dual-claim RNA assay is a plus for patient care, Masciotra said. “However, additional factors, such as the implications of an off-label use and costs associated with the implementation of a second-step NAT algorithm, need to be explored.”

Amy Carpenter Aquino is CAP TODAY senior editor.