The scale of that undiagnosed population underpins the urgency of the work. According to UNAIDS, roughly 40.8 million people were living with HIV worldwide in 2024, with 1.3 million new infections reported annually, and an estimated 5.3 million people remain unaware of their infection. China, with 1.35 million people living with HIV and about 100,000 new cases in 2024, has reached 84 percent of the first UNAIDS “95” target, meaning approximately 200,000 infections are still undiagnosed. Late presentation compounds the problem: about half of people newly diagnosed globally, and more than 60 percent of those diagnosed in China between 2013 and 2023, arrive at care with CD4+ T cell counts below 350 cells per microliter or with AIDS-defining illnesses, outcomes that raise transmission risk, morbidity, and mortality.

The diagnostic pathway responsible for confirming infection in China still leans heavily on serology. A reactive antibody screening test is typically followed by a Western blot confirmation, a technique that detects antibodies against individual HIV proteins but carries a median window period exceeding 40 days, far longer than the roughly 11.5-day median window of nucleic acid testing. Nationwide reports suggest that nearly 10 percent of reactive screening samples yield indeterminate Western blot results, and most of those individuals are ultimately confirmed HIV-positive through supplemental testing. The United States CDC revised its own algorithm in 2023 to incorporate an HIV-1 nucleic acid test as a confirmatory step, but China has lacked direct clinical evidence for making the same shift. The new study, led by researchers at Beijing Youan Hospital of Capital Medical University in collaboration with Southeast University, set out to provide it.

Between June 2024 and May 2025, the team enrolled 215 adults presenting for first-time HIV testing at Beijing Youan Hospital, excluding anyone reporting prior antiretroviral therapy or pre- or post-exposure prophylaxis. The cohort comprised 113 HIV-negative and 102 HIV-positive individuals, with infection status established by a composite gold standard integrating laboratory results, clinical findings, and epidemiological history; all participants classified as negative were followed for at least three months to exclude acute infection. Each blood sample was split and tested in parallel on four platforms: an ELISA detecting antibody and P24 antigen, quantitative HIV-1 RNA measurement on the Abbott RealTime system, a qualitative multiplex RNA PCR, and a quantitative proviral DNA assay run on lymphocyte-enriched cells and normalized to copies per million nucleated cells.

The performance hierarchy that emerged was striking. Quantitative DNA testing correctly identified all 102 positive and all 113 negative samples, yielding 100 percent sensitivity, 100 percent specificity, and an area under the ROC curve of 1.000. RNA quantification came close behind, missing a single positive sample for 99.02 percent sensitivity with perfect specificity and an AUC of 0.995. ELISA achieved the same 99.02 percent sensitivity but produced two false positives, dropping specificity to 98.23 percent. The qualitative RNA assay matched the others’ sensitivity but generated 28 false positives among the negative samples, giving it a specificity of just 75.22 percent and an AUC of 0.871, significantly inferior after Bonferroni correction. Notably, sensitivity remained consistent across all CD4+ T cell strata for every method, an important property for detecting infection regardless of immune status.

The two diagnostic misses tell instructive stories. The sole false negative shared by both RNA assays came from a patient whose CD4 count exceeded 700 cells per microliter and who later admitted to a month of antiretroviral treatment before sampling despite initially denying it; suppression of viral RNA by effective therapy drove the result below the detection limit. Proviral DNA, by contrast, stayed positive, because the integrated viral genome persists in host cells irrespective of plasma viral load. Viral RNA is also fragile in pre-analytical terms, degrading readily when plasma separation or cold-chain transport is delayed, while proviral DNA is far more stable. The single ELISA false negative occurred paradoxically in a patient with a very high viral load and a CD4 count of 155 cells per microliter, a pattern the authors attribute to the Hook effect, in which excessive antigen swamps antibody binding sites in the sandwich assay, compounded by B-cell dysfunction in advanced immunosuppression.

To translate individual assay performance into workflow recommendations, the researchers simulated serial tandem testing strategies. The optimal combination was ELISA screening followed by DNA quantitative confirmation, which achieved a net sensitivity of 99.02 percent, net specificity of 100 percent, and required 318 total tests across the cohort. Algorithms beginning with nucleic acid screening matched that sensitivity and specificity but demanded between 316 and 344 tests. The authors emphasize that at their institution, the patient-end cost of DNA quantification is comparable to RNA viral load testing, making a targeted ELISA-plus-DNA confirmation step a practical and affordable upgrade rather than a financial burden, particularly for vulnerable populations.

Perhaps the most clinically consequential results came from the 18 samples with indeterminate Western blot findings, ten from people ultimately classified negative and eight from confirmed positives. On these difficult cases, both DNA and RNA quantification achieved 100 percent diagnostic accuracy, resolving every ambiguity the traditional confirmation test could not, while qualitative RNA testing reached 94.44 percent accuracy and ELISA only 83.33 percent. Western blot indeterminacy typically arises from incomplete antibody responses in early infection or from nonspecific cross-reactivity, and NAT sidesteps the problem entirely by detecting viral nucleic acids within roughly 3 to 5 days of infection, independent of the host immune response. Earlier work has shown that more than 99 percent of people with indeterminate Western blot results and detectable viral load are eventually confirmed positive, and the new data extend that logic to DNA as a marker immune to antiretroviral suppression.

The study’s authors argue that their findings provide the direct evidence China has lacked for embedding nucleic acid testing in its confirmation algorithms, and they see DNA quantification as especially valuable in two scenarios: resolving indeterminate Western blots and diagnosing people with undisclosed antiretroviral exposure, a growing issue as pre- and post-exposure prophylaxis and informal treatment use expand. The same stability that makes proviral DNA reliable in adults makes nucleic acid testing indispensable in infants, where maternal antibodies can persist for up to 18 months and render serology uninformative, and in pregnant women who need rapid confirmation to start mother-to-child prevention. Emerging technologies, including CRISPR-Cas12a systems coupled with nested PCR, continue to push nucleic acid testing toward higher sensitivity and greater accessibility.

The researchers are candid about the limitations. The single-center cohort was overwhelmingly male and dominated by homosexual transmission, so multicenter studies with greater demographic diversity, including pregnant women and newborns, will be needed to confirm generalizability. Excluding acutely infected and treated patients may have inflated real-world performance estimates, and the analysis did not formally model turnaround time, operational complexity, or cost-effectiveness in primary care. Still, the core conclusion is difficult to escape: a first-generation DNA quantification kit, only available since September 2024, matched clinical diagnosis perfectly in this cohort where serology and RNA assays each stumbled at least once. If validated more broadly, an ELISA-first, DNA-confirm algorithm could shorten diagnostic windows, cut late diagnoses, and move national AIDS programs measurably closer to the first 95 percent, the hardest link in the 95-95-95 chain.

Subject of Research: Comparative effectiveness of nucleic acid and serological assays for optimizing HIV diagnostic algorithms in clinical practice

Article Title: Nucleic acid testing-optimized HIV diagnostic algorithms: a multimethod comparative effectiveness analysis in clinical practice

Article References: Nucleic acid testing-optimized HIV diagnostic algorithms: a multimethod comparative effectiveness analysis in clinical practice. (n.d.). https://doi.org/10.3389/bjbs.2026.15887

Image Credits: AI Generated

DOI: 10.3389/bjbs.2026.15887

Keywords: HIV diagnosis, nucleic acid testing, proviral DNA quantification, ELISA, Western blot, viral RNA, 95-95-95 targets, late diagnosis, diagnostic algorithms, Beijing Youan Hospital, China, comparative effectiveness