Large US study shows which HIV tests are most accurate

Antibody-only tests need to be replaced with combination tests

Roger Pebody
Published: 06 January 2014

Differences in the performance of commonly used HIV tests lead to substantial differences in the number of infections which are diagnosed in everyday clinical practice, according to an analysis from San Francisco published last month in PLOS ONE. If the test used only detects HIV antibodies, most acute (recent) HIV infections are missed, with the OraQuick rapid test also missing established infections when testing saliva samples.

However, a laboratory combination test – which can detect both antibodies and p24 antigen – has excellent performance, including in people who have acute infection. While a rapid combination test did not have as good results as the laboratory version, it was superior to rapid antibody tests.

In the first part of the study, clinicians re-examined the performance between 2003 and 2008 of a range of HIV tests used with high-prevalence San Francisco populations (sexual health clinics, a clinic for sex workers, people seeking PEP, people testing during contact tracing).

During this time in the city, antibody laboratory or rapid tests were supplemented with pooled testing of HIV RNA (i.e. a viral load test). This is considered the gold standard test for acute HIV infection – a person who tests negative on an antibody test but positive on HIV RNA is likely to have acquired HIV in the last few weeks. It is important to diagnose individuals who have acute infection as they have exceptionally high viral loads and contribute disproportionately to onward HIV transmission.

One aim of the study was to assess possible alternatives to the use of HIV RNA testing, as it is expensive and complicated to perform. For the second part of the study, the researchers therefore took stored samples of blood plasma of 58 people who had been diagnosed with acute infection and re-tested them using a wide range of tests, including fourth-generation ‘combination’ tests.

Across the 21,234 tests that had been performed, 761 people had been diagnosed with HIV (prevalence 3.6%), including 58 people with acute infection (prevalence 0.3%).

The main results are in the table. The key measures of accuracy are sensitivity (the percentage of results that are correctly positive when HIV is actually present) and specificity (the percentage of results that are correctly negative when HIV is not present).

As shown by the specificity scores of 99.9% or above, the study provides reassurance that false positive results are exceptionally rare – in other words, people who are HIV negative do correctly receive HIV-negative results. 

Type of test

Product

Specificity (correct negative results), based on 21,234 tests

Sensitivity (correct positive results), based on 21,234 tests

Sensitivity for acute infection only, based on retesting 58 samples

Estimated overall sensitivity

Lab antibody (1st gen)

Vironostika

100%

92.3%

0%

92.4%

Lab antibody (3rd gen)

Genetic Systems

100%

96.2%

34.5%

95.0%

Rapid antibody (3rd gen), saliva

OraQuick Advance

99.9%

86.6%

-

-

Rapid antibody (3rd gen), blood

OraQuick Advance

100%

91.9%

5.2%

92.8%

Rapid antibody (3rd gen)

Uni-gold Recombigen*

-

-

25.9%

94.3%

Lab antibody + p24 antigen (4th gen)

Architect

-

-

87.3%

99.1%

Rapid antibody + p24 antigen (4th gen)

Determine

-

-

54.4%

96.6%

* Results for two other rapid antibody tests are not shown here, but were broadly similar.

As expected, the ‘first-generation’ laboratory antibody test (first introduced in 1987) delivered a number of false negative results to individuals who did have HIV (sensitivity 92.3%), especially individuals with acute infection.

The ‘third-generation’ laboratory antibody test performed better, with a sensitivity of 96.2%. But rapid versions of antibody tests were not as sensitive, with results similar to those of the older laboratory test.

In particular, the OraQuick test (a version of which has been recently licensed by the Food & Drug Administration (FDA) for self-testing) had poor results when testing samples of saliva (rather than samples of fingerstick blood). All individuals with acute HIV infection and one-in-twenty individuals with established HIV infection were given false negative results by this test. This is not the first study to observe such problems. “These concerns must therefore be carefully weighed against the potential advantages of home based testing or oral fluid testing when choosing a testing strategy for individuals at high-risk of HIV infection,” comment the authors.

By far the best results came from a ‘fourth-generation’ laboratory test. Although such tests have been used in Europe since the late 1990s, it took until 2010 for one of these tests to be approved by the FDA for use in the United States. As fourth-generation tests combine detection of antibodies with detection of p24 antigen (levels of which increase soon after infection, before antibodies are produced), they are better at diagnosing people during acute infection than third-generation tests.

In this study, 87.3% of people with acute infection, and an estimated 99.1% of all people with HIV in this setting, would be correctly diagnosed with a fourth-generation laboratory test. Furthermore, additional tests on 81 HIV-negative samples suggested a specificity of 100%.

The researchers also tested a ‘fourth-generation’ rapid test. This device delivers results within minutes, while having the unique selling point of detecting both antibodies and p24 antigen. However, as has been reported in previous studies, its detection of acute infection is not as accurate as laboratory ‘fourth-generation’ tests.

On this occasion, just over half the individuals with recent infection were correctly diagnosed (sensitivity 54.4%). The authors say that the test was able to diagnose those individuals with viral loads above 500,000 copies/ml, but not below this level.

Overall, the researchers estimate that 96.6% of all people with HIV in this setting would be correctly diagnosed with this test. While this result was poorer than for the laboratory test, it was better than for any other rapid point-of-care test that was examined.

The authors note that their findings are mostly relevant to settings such as San Francisco where the prevalence of acute HIV infection is high. In lower-risk settings, this will not be such a concern.

“Results suggest that the availability of newer, 4th generation combo immunoassays or point-of-care rapid tests could represent a major advance for HIV diagnostics and prevention by permitting much more rapid testing for acute HIV infection in diverse clinical settings,” they conclude.