Weighing the risks in HIV, HCV algorithmic testing

The tiebreaker molecular testing isn’t performed automatically. “It’s orderable as a standalone test for our providers. Prior to 2016, this test could simply be added on to an in-lab specimen. A provider could see that it was needed, they would call the laboratory, we’d pull the specimen, and we would run it on the same specimen on which we ran the serologic testing.” Physician compliance was 100 percent, based on a review of molecular testing from 2014 to 2016, with all molecular results reported within 10 days of the reactive Ag/Ab screen. “The vast majority of our patients were tested the same day. So overall this was reassuring,” he said.

In 2016, considering evidence that specimen-to-specimen carryover may result in an erroneous diagnosis of HIV, a prevention policy was implemented. “The decision was made to cease doing any sort of molecular testing for HIV, HCV, or HBV on specimens other than dedicated specimens that were sent to the lab for that testing.”

A one-year comparison of the data pre- and post-policy found that all patients who needed follow-up molecular testing received it. “The bad news was that we did have two patients with significant delays in follow-up testing—delays of almost half a year,” and one of those patients had HIV. “So that’s what we were balancing. In approximately one year’s time we had a couple of patients who did not receive a timely diagnosis, versus every year or so—estimated based on the scale of our testing—we might have a false-positive that could lead to a misdiagnosed patient. Based on that analysis we decided we wanted to keep the practice we have, and we still require a dedicated tube for additional molecular testing for HIV.”

Once the risk of delay became known, they implemented processes to mitigate the risk of loss to follow-up. Technologists now call the differentiation assay result to the physician who ordered the original test, following a script that emphasizes the need for an additional sample. The script is also added as a comment to the patient result. In addition, a daily report is compiled of all differentiation assay results and is made available to select HIV specialized providers. “There have been instances of patients they’ve tracked down after the fact because they realized they had a negative differentiation assay and needed that tiebreaker molecular testing.”

“So we have some stopgaps.”

HCV confirmatory molecular testing is also offered as a standalone test at Barnes-Jewish, rather than performed automatically. When Dr. Anderson and colleagues reviewed compliance with testing, they found that from 2010 to 2016, only 48 percent of HCV antibody-reactive specimens (1,329/2,792) received molecular follow-up within 30 days. “So less than half of patients were receiving testing to figure out whether they had active disease,” he said. And of those who were serologically positive and did indeed receive molecular testing, only three quarters (2,715/3,607) had active infection, with the remainder false-positives or cases of cleared HCV. “If providers are actually using the serologic result alone to diagnose HCV, we’re going to have a problem.” Though similar to the challenge with HIV follow-up testing, given the comparative amount of patients who need HCV molecular testing, “the problem is far more significant,” he said.

One option is to perform HCV molecular testing on a separately collected sample. This was their approach for a long period and is coupled with a comment that strongly emphasizes testing is incomplete and additional molecular testing is needed. “That’s what we did for a long time—it’s similar to what we do with our HIV testing.” But by the time physicians saw the comment and would have ordered the molecular test, many of the patients seen in the ED were no longer available for the second draw.

A second option is to collect two specimens from each patient in case confirmatory testing is needed. The Barnes-Jewish laboratory performs about 200 HCV antibody tests a day, with 14 percent positivity. “So we’d be looking at about 180 tubes of plasma a day that we would be saving and eventually be discarding,” Dr. Anderson said. “It would be a lot of wasted time and space.” That leaves reflex testing, performed on the same tube. “And once again we’re back to balancing that risk of incomplete testing versus contamination. But this is different from HIV—it’s more dire. We know a lot of this reflexive testing isn’t happening.”

The literature on risk of HCV RNA contamination isn’t fully consistent, he said. In one study, Rondahl, et al., tested for contamination known HCV RNA-positive and RNA-negative samples (149 of each) in an alternating fashion by the Abbott anti-HCV assay in an Architect instrument. In subsequent retesting of the previously RNA-negative samples, six of the 149 were positive by the Roche Cobas TaqMan assay (Rondahl E, et al. J Clin Virol. 2014;60[2]:172–173). “What they found was a four percent carryover rate,” Dr. Anderson said. “That’s pretty astounding.” The maximum viral load in the contaminated samples was 33 IU/mL. “So very low viral loads, which is important.”

A larger study investigated the potential risk of HCV, HBV, and HIV nucleic acid cross-contamination on 480 negative specimens by a serology screening instrument that uses disposable tips for sample transfer before molecular testing. The negative plasma samples were subsequently tested with the Cobas HCV test, the Cobas HBV test, or the Cobas HIV test on the Cobas 6800. The authors found no evidence of cross-contamination in the 480 negative specimens on the serology module (Cobas e 602) after processing alongside of high-titer HCV, HBV, and HIV-spiked specimens (Rodriguez PL, et al. Sex Transm Dis. 2020;47[5S]:S32–S34).

“So what can we make of this?” Some instruments could have more risk than others, Dr. Anderson said. The 2014 study used serology screening instruments with a fixed needle for sample transfer; the 2020 study used the instrument with the disposable tip. “But one thing I would argue, as any of us in the clinical lab know, is it’s more than just the instruments that could be contributing to this. How the tubes are handled, if there’s any aliquotting steps, if they’re going on an automated line where they’re spun ahead of time—all of that could contribute.” When it’s multifactorial, Dr. Anderson added, “you need to do your own study.”

That’s what the Barnes-Jewish laboratory did. They prepared 10 negative contrived specimens from pooled serum from HCV-negative patients and 10 positive contrived specimens from negative pooled serum spiked with deactivated HCV control material to a concentration of 2.15×10⁷ IU/mL. Aliquots from each pool were prepared into serum separator tubes, he said, “to mimic a real-world setting as much as possible,” and numbered one to 20, alternating positive and negative. Sets of the same 20 alternating samples were distributed to the BJH core serology laboratory and the serology laboratories of two sister hospitals that send molecular testing to BJH. The labs were instructed to follow the normal process for serology testing, while maintaining the one to 20 order, after which the specimens were sent to the BJH molecular laboratory for testing. “We tried to blind them to the reason we were doing this.”

Hospitals one and two performed serologic testing on the Abbott Architect, and hospital three used the Roche Cobas 8000. All molecular testing was done by the Cobas CAPTAQ HCV test on the TaqMan. Hospitals one and two had a zero percent contamination rate, with 10/10 samples negative by the CAPTAQ assay. The third hospital had a 10 percent contamination rate, with a single positive sample. “And it had a viral load of 21 IU/mL,” Dr. Anderson said. “I interpreted this to mean that carryover contamination is indeed a risk in our hospital system. We can dissect this and try to figure out what we can do to mitigate the risk, but we need to accept that it can happen. However, there’s something interesting”: In the BJH study and in the others, the specimens with carryover contamination almost universally have very low viral loads. “And that’s something we can work with, so we did implement a reflex algorithm” (Fig. 1).

In that algorithm, if serologic testing is reactive, it’s reported as positive, “and we tell our providers we’re going to perform reflexive molecular testing.” A separate order, generated in Epic, that’s unique to the reflex test is used. “We call it our reflex HCV RNA.” If it’s detected, with a viral load of more than 1,000 IU/mL, it’s reported as detected, with the value. If it’s detected and the value is less than 1,000 IU/mL, “there’s a chance it could be contamination,” he said, in which case it’s called indeterminate, the value is not reported, and repeat testing on a new dedicated specimen is recommended. “We do see this happen occasionally with this reflex. I don’t know if those patients are infected or not, but it’s something we’ve used successfully. And it means the vast majority of our patients get the full testing.”

HCV RNA testing not linked to the reflex algorithm can be ordered in Epic. “That does require a dedicated tube, and for that we would report values less than 1,000.” This is important, he said, because in the setting of monitoring a known positive patient, values less than 1,000 regularly occur and are relevant.

A “major lesson” learned through this process, Dr. Anderson said, is that when designing algorithmic testing, collaborating with all stakeholders is a must, as is sizing up the full testing process. “You can validate each component of the assay, you can do all the carryover studies you want, but if you’re not taking into account the entire testing process, you’re not capturing the risk.” 

Charna Albert is CAP TODAY associate contributing editor.