Amy Carpenter Aquino
March 2020—Where should molecular infectious disease testing be performed?
That was the question at the center of a point-counterpoint session at last year’s Association for Molecular Pathology annual meeting, where two speakers debated who owns molecular infectious disease testing: the clinical microbiology laboratory or the molecular pathology laboratory?
“This isn’t my first rodeo,” said Frederick Nolte, PhD, D(ABMM), professor and vice chair for laboratory medicine, Department of Pathology, Medical University of South Carolina. In 2012, he defended the centralized laboratory approach (Mosammaparast N, et al. J Clin Microbiol. 2012;50[6]:1836–1840). Eight years later, he defended the distributed molecular testing model (Longshore JW. Clin Chem. 2020;66[1]:138–139; Nolte FS. Clin Chem. 2020;66[1]:140–142). “Some tests belong in microbiology, some belong in the HLA lab, some belong in molecular pathology,” he said at the meeting.
Broad consolidation of molecular diagnostics can result in consolidated and standardized workflows, test processes, and platforms, which lead to greater efficiency and lower costs. “As we’re all aware, the pool of skilled technologists with molecular skills is not limitless,” Dr. Nolte said, “so broad consolidation of molecular diagnostics provides for cross-training and utilization of those technologists to the greatest extent.”
A large consolidated laboratory offers the opportunity to institute multiple shifts, which can result in shorter turnaround times, and the unique lab designs required for many molecular pathology tests need not be replicated throughout the clinical laboratory.
Decentralization of molecular diagnostics is driven primarily by technology. “The simplification of the enabling technologies, highly automated sample-in, answer-out platforms, as well as point-of-care tests based on nucleic acid amplification are a reality,” Dr. Nolte said, noting that labs are less reliant now on laboratory-developed tests.
“Molecular diagnostics is now well integrated into the workflow of many areas in terms of laboratory medicine,” he said. “The results often complement the traditional tests that are being done,” which presents the opportunity for “diagnostic integration.”
At MUSC, molecular diagnostics testing is distributed. The microbiology laboratory runs respiratory, GI, and blood culture identification panels on the BioFire FilmArray, and enterovirus, group A Streptococcus, C. difficile, and Mycobacterium tuberculosis assays on the Cepheid GeneXpert. “These tests are run on demand,” Dr. Nolte said, “and since the lab is staffed 24/7, we can offer rapid turnaround time for results.”
The molecular pathology laboratory offers a wider menu of laboratory-developed tests for infectious diseases and some that fall outside the ID testing arena, Dr. Nolte said. “It’s a multidisciplinary molecular pathology lab but heavily weighted toward infectious diseases,” he said. The molecular and cytogenetics laboratory runs FISH, SNP microarrays, and solid tumor and myeloid NGS panels, among other tests. Both labs have a single shift.
“We work very well together, and we shift resources back and forth between labs as the technology changes and needs arise,” Dr. Nolte said.
[dropcap]S[/dropcap]aying that he let Dr. Nolte lead the introduction because it would be his “only win of the day,” Nathan A. Ledeboer, PhD, D(ABMM), professor and vice chair, Department of Pathology and Laboratory Medicine, Medical College of Wisconsin, said the debate centered on a few questions. The first: “Infectious disease molecular diagnostic platforms are increasingly automated and include broad test menus. Are tests still ‘owned’ by a laboratory section?”
“In other words, should we continue to be siloed laboratories?” Dr. Ledeboer said.
Dr. Nolte began by saying that the latest generation of molecular diagnostic platforms for infectious diseases are completely automated, high-throughput instruments offering random access and designed with total laboratory automation in mind. The three key systems—from Roche, Hologic, and Abbott—can be deployed in a multidisciplinary molecular pathology laboratory, in a microbiology lab, or even in a core chemistry lab. “They require no special expertise in microbiology or molecular biology to operate,” he said.
Nucleic acid amplification testing has been incorporated into diagnostic algorithms for hepatitis C and B viruses and HIV, for which serological tests are typically performed in the core chemistry laboratory. “This offers the possibility of true reflexive testing” from the same sample used for serology, he said. While there are hurdles—different sampling requirements and environmental contamination and immunoassay instrument specimen carryover risks—“they’re not impossible barriers to work through.”
Yes, labs are increasingly using NAATs for HIV, HCV, and cytomegalovirus diagnosis, Dr. Ledeboer said. “The challenge is when we’re trying to make a diagnosis, we need to be able to report an absolutely correct result the first time. This is especially important as you start to think about how data is transmitted into the medical record.”
Providers and patients often receive the test results simultaneously, Dr. Ledeboer said, and a low-positive result caused by contamination can be extremely concerning. “If we report low-level positives, even with comments, that can be confusing to the providers and our patients. We need to absolutely make sure that our testing is done in an accurate manner” (Bryan A, et al. Clin Chem. 2016;62[7]:973–981).
Dr. Ledeboer said that while he loves the idea of connecting a molecular instrument to a chemistry or hematology line, the drawback is that high-need analytes may take priority over infectious disease testing. “Chemistry and hematology lines assign various priorities to individual tests,” he said. “While an HIV or a CMV from a transplant patient may be very high priority to us in the microbiology laboratory, that may be much lower priority compared to a troponin on a patient with an acute MI, and that may result in significant delays in turnaround time.”
Another hurdle: Core laboratory lines are largely suited for blood-based systems. “We perform testing on multiple different specimen types” in microbiology and molecular diagnostics, Dr. Ledeboer said, and that “presents a very significant challenge, especially in a consolidated setting.”
Dr. Nolte disagreed, saying the instruments were designed to handle different specimen types with automation solutions to match their test menus in different specimens. “I don’t see that as a particular hurdle,” he said. While it is true that the core chemistry laboratory is more accustomed to receiving body fluids and blood than stool, “they can get over it.”
Dr. Nolte also disagreed with Dr. Ledeboer’s argument about priorities and delayed turnaround times. Yes, HIV, HCV, and other such tests might not have the same priority as other core lab tests, Dr. Nolte conceded, but “the turnaround time would certainly be an improvement over shuffling specimens from one laboratory to another, particularly if you were talking about things like HIV viral load testing, hepatitis C, and HPV. They’re batched tests now for most laboratories that are doing the tests.”
Of the low-level contamination risk, Dr. Nolte said, “There’s no denying that a core chemistry lab has [risk for] environmental nucleic acid contamination,” but so do the microbiology and molecular pathology laboratories. While agreeing it’s a problem that has to be addressed, he said, “it’s unfair to single out the core chemistry lab.”
Dr. Nolte shared an MUSC study on clinical validation of reflexive hepatitis C virus RNA testing directly from HCV antibody reactive specimens.
“We have increased our efforts in our emergency departments to screen for hepatitis C, such that we’ve gone from testing in the neighborhood of 7,000 specimens per year to 27,000 specimens,” Dr. Nolte said. “With that comes the need to link those patients with active HCV infection to appropriate care. Linking the serology to the RNA testing seamlessly was a big concern.”
His team considered various testing models. “Many laboratories that have instituted these reflexive testing algorithms ask for two specimens up front: one that they hang on to until they know the antibody results and then use for the nucleic acid test. That works, but it’s a lot of specimen management, and depending upon your HCV antibody positivity rate, most of those tubes of blood are going to be wasted.”
In anticipation of studying the streamlined approach of using one specimen for antibody and RNA testing, “we did some homework,” Dr. Nolte said. They use the lab’s Abbott Molecular RealTime HCV assay for HCV RNA testing and the Abbott Diagnostics Architect i2000 immunoassay analyzer for HCV antibody testing. “The problem was that the Architect instrument uses a single probe to sample every specimen. It’s washed 10 times between samples, but it’s the same probe,” so there’s specimen carryover risk.
They asked Abbott Diagnostics about carryover contamination. “Not a problem,” they were told, “because they were used to talking about carryover of antibodies or antigens,” Dr. Nolte said. He told Abbott his team was talking about putting PCR in back of the immunoassay. Abbott’s response: There is no data.
Now there is. The MUSC study investigated the risk for carryover of HCV RNA for samples processed for HCV antibody testing in the clinical chemistry lab to determine whether the risk warrants collection of a separate specimen for reflexive RNA testing. They found that while HCV RNA carryover does occur with the i2000 immunoassay analyzer, it is infrequent and at very low levels (presented at ASM Microbe 2017).
“We first challenged the immunoassay analyzer with 100 positive serum specimens that had greater than one million copies or one million international units per milliliter of HCV,” Dr. Nolte said, and followed each one with a negative serum specimen. “That’s a pretty significant challenge. We did find contamination, there was some carryover.” More than five percent of the 108 negative samples were cross-contaminated, but all six false-positive samples contained low levels of HCV RNA (<1.08 log IU/mL). “That’s manageable,” he said.
The laboratory manages the risk with a conservative limit, he said. “If we do a reflexive test like this—if we get anything less than 1,000 copies per milliliter—we request a second sample,” Dr. Nolte said. “That only happens two or three percent of the time.” Similar results were reported in the Journal of Clinical Virology (Rondahl E, et al. 2014;60[2]:172–173).
His team did not conduct environmental sampling but did run several hundred specimens through the complete procedure. “We tested all those antibody-negative specimens for HCV RNA,” he said. “None of them was positive.”
The MUSC laboratory is still performing separate points of service, he said. Antibody testing is done in the core chemistry laboratory, with all positive results reflexed to the molecular pathology laboratory for HCV RNA testing. “It has done a great deal to facilitate the diagnosis and to link those patients to care, particularly coming from the emergency department,” he said, adding, “This is a nice jumping-off point to think about moving the testing site for hepatitis C RNA to our core chemistry lab.”
[dropcap]W[/dropcap]ith testing moved out of the microbiology lab or, in the case of MUSC, the molecular pathology lab, Dr. Ledeboer said, the questions become: “Who’s going to help interpret these results? And ultimately, who will flag the issue that this is a patient of concern?”
He argues that the microbiology laboratory is best positioned to draw a clinician’s attention to emerging resistance in a patient with elevated viral loads. “This is a function we actively perform in the microbiology laboratory today because we see results, we review results, and we make sure to look at who’s continuously positive on a weekly basis,” Dr. Ledeboer said.
Troubleshooting, too, requires expertise and is “incredibly complicated,” he said, referring to the problem of contamination and citing a lengthy list of steps to use in the presence of possible contamination (Greub G, et al. Future Microbiol. 2016;11[3]:403–425). “These are things that microbiologists and molecular biologists are well trained to do and think about.”
Molecular infectious disease testing is becoming more efficient, thanks in part to automation, and on that point there was no disagreement. “Regardless of what methodology we choose, we’re becoming more efficient,” Dr. Ledeboer said. The number of specimens and the number of test requests for molecular infectious disease testing are increasing while staffing is stable (Greub G, et al. Future Microbiol. 2016;11[3]:403–425). “Even though we’re expanding the amount of testing, regardless of where testing is being performed, we’re ultimately being more efficient as a laboratory in general,” he said.
“I can’t agree more,” Dr. Nolte added. His lab hasn’t added staff in 12 years despite yearly workload increases of 10 percent or more. “We have managed to keep ahead of it because of the increased efficiency,” he said.
Another area of agreement: C. difficile testing is best performed in the microbiology lab, where specialty expertise drives the correct use of testing algorithms. Said Dr. Ledeboer: “We may use NAAT as a screening test or for certain locations, depending on prevalence and pretest probability of disease—NAAT alone versus a NAAT with an algorithm. Using these types of algorithms, we have taken advantage of the expertise we have in the microbiology lab in helping to interpret these, particularly as it relates to our physicians calling when they get a NAAT-positive, EIA-negative result with an interpretation of it being indeterminate” and the need to look for another source of the patient’s diarrhea. “This is where our microbiology staff can begin to shine and help us as we start to interpret these questions.”
Dr. Nolte agrees: “C. difficile is the sweet spot for our microbiology lab as well. When it was a batch test, we had it in the molecular pathology lab because at the time it was an LDT and it needed a little more care and feeding than the Cepheid [GeneXpert] platform.”
In addition to the microbiology and molecular pathology labs, “it seems we have a third player,” said moderator Esther Babady, PhD, D(ABMM), director of microbiology and molecular microbiology at Memorial Sloan Kettering Cancer Center, referring to the core chemistry lab. “What does chemistry have to do with any of it? Is there molecular testing in the chemistry lab?”
Abbott and Roche instruments were designed with total laboratory automation in mind, Dr. Nolte replied, “and that usually means the core chemistry laboratory, the stat laboratory, if you will.” They are thinking about that as a point of service for many of the high-volume molecular tests. “So that’s something we’re going to have to consider.”
“I completely agree,” Dr. Ledeboer said. The Roche Cobas 8800, for example, is designed as a random-access instrument that can perform viral load testing on demand or near on demand, “so we’re getting quite close to being able to perform this testing in the clinical core and the chemistry labs.”
“It’s definitely a place that’s very much coming.”
[dropcap]T[/dropcap]hey tackled another ownership question: “As the role of sequencing has broadened, how do we consider the cost of instrumentation and the role of bioinformatics in who owns testing?”
Dr. Nolte said metagenomic next-generation sequencing is likely the future state of clinical microbiology, citing a recent article (Simner PJ, et al. Clin Infect Dis. 2018;66[5]:778–788). “It honestly could possibly do everything we’re doing by conventional methods, do it on a timeline that’s consistent with patient care, in one to two days. But there are an awful lot of hurdles to be overcome before we think about this being the future state of the clinical microbiology lab.”
Dr. Ledeboer has a different view. “Microbiology laboratories pioneered the LDT” for respiratory viruses, MRSA, and viral load testing. “We are innovative and can do this in the micro lab. We know how to perform extraction and DNA sequencing,” and library preparation can be automated and learned. Yes, there is bioinformatics and there are databases, he said, “but at the end of the day, so much of what we need to think about when we consider sequencing is the right test for the right patient at the right time—and we’re the ones who can deliver that.”
He shared the health care system’s sequencing data studied by a colleague. “He looked at what is useful of our 16S sequencing that we send out,” Dr. Ledeboer said.
Of 163 specimens, 4.3 percent (seven patients) were sequencing-positive, culture-negative because the patient was on an antibiotic; the sequencing ultimately delivered clinical value. “This is guidance that only a trained microbiologist can provide,” Dr. Ledeboer said. And “we’ve been well trained to assess for contamination.”
What about interpretation? A patient presented to Froedtert Hospital and the Medical College of Wisconsin with Chagas disease. After struggling and waiting for a diagnosis, “we sent a plasma-based Karius test on this patient that came back and gave us a diagnosis within 24 hours for Trypanosoma cruzi.”
The patient’s test results were positive for several other microorganisms. “The resident and ID physician who were caring for this patient called me and said, ‘What do we do with this? What does it mean? I think I’m going to ignore it.’” They believed the Chagas but weren’t worried about the rest, and Dr. Ledeboer told them he tended to agree.
But Dr. Ledeboer suggested relooking at the patient and the imaging. “We found that the patient had a perforated bowel that was leaking GI flora into the abdominal cavity.”
Consultations with various teams resulted in finding the right information to help the patient, who did have Chagas, he said. “This is largely why we need to be able to do this and maintain this expertise in the microbiology laboratory.”
Dr. Nolte listed the barriers to implementing metagenomics in the microbiology laboratory:
- Equipment costs (“but they’re coming down”).
- Need for data storage and analysis and databases, as well as bioinformatics and variant scientist support. “We can barely keep up on our cancer genomic side in terms of having enough support.”
- Nucleic acid contamination. “It’s not only a concern where the test is done. It can be introduced during specimen collection” and at other steps in the process.
- Metagenomic assays for infectious disease are currently LDTs and their validation and verification add a new level of complexity compared with targeted molecular techniques. “It scares me to think I would have to validate one of these tests.”
- mNGS assays are currently offered as billable LDTs by clinical and commercial labs, but clinical utility is still under investigation and the assays don’t replace standard methods. “They tend to be a method of last resort, which is probably not the best way to use them. It offers tremendous opportunities for us in terms of laboratory stewardship—no argument there.”
- Interpretation is one of the greatest challenges, and “you don’t lose as a microbiologist in this. You want to be part of the team that helps in interpretation. But you don’t own it all. I don’t own it all. . . . We’re major players, but we don’t own it.”
[dropcap]P[/dropcap]oint-of-care testing, too, was on the table in this debate: “Increased breadth of the point-of-care testing menu may mean less volume for the laboratory. How does the laboratory continue to have a voice when testing is performed outside the lab?”
The laboratory doesn’t lose its voice, Dr. Nolte said, so long as it’s engaged in the decision to implement it and lab personnel are on the POC team to ensure the necessary procedures are followed.
But point of care is going to take testing away from the laboratory in general, Dr. Ledeboer said. “When we hear requests for point-of-care testing, it’s not because the laboratory test is bad or doesn’t deliver accurate information. It’s because at the end of the day, we’re not providing results in the appropriate turnaround time.”
“We need to focus our point-of-care work on where the laboratory can’t meet the needs rather than look at it as an alternative to performing testing in the laboratory as a whole,” he said.
New CAP checklist requirements released last fall mean POC testing “loses most of its exceptions where CAP accreditation is used,” Dr. Ledeboer said, citing POC.08675 Quality Monitoring Statistics, POC.08690 Specimen Handling Procedures, and POC.08715 Safe Specimen Handling/Processing. “We are all going to be held to the standard, regardless of where testing is performed, of doing quality monitoring statistics. This is what we already do in microbiology and molecular pathology. We’re used to this.” If labs can meet clinicians’ needs, he said, “we can eliminate having to worry about that requirement” and the others—and provide more reliable results.
Dr. Nolte agreed with Dr. Ledeboer’s take on POC testing (“it does create significant challenges”), but he said it supports his view that the laboratory is going to “let go” of a number of its tests to be performed as POC tests. “There are a number of things that have to be addressed for that to happen effectively,” he added.
At MUSC, Dr. Nolte was happy, he said, to hand over POC testing to Nina Babic, PhD, director of the clinical chemistry laboratories and POC testing. “Nina and I are working together to try to figure out how we might deploy molecular assays for infectious diseases as point-of-care tests.” And the questions are many: “In the emergency department, where it’s a 30-second ride in the pneumatic tube system from the ED to the clinical lab, do we do rapid influenza tests there? Do we do point-of-care tests in a core chemistry lab? Or another type of test in the clinical microbiology lab? Is it focused? Is it a broad panel? How do you balance all of this? That is the biggest challenge,” Dr. Nolte said, urging discussions with point-of-care test coordinators. “They desperately need your expertise,” he said, “not only in terms of microbiology but in terms of molecular diagnostics.”
While there is about a 20 percent difference in sensitivity between rapid antigen detection tests and NAATs for influenza, at least with regard to testing for influenza A, Dr. Nolte asks: “Is the juice worth the squeeze?” NAATs could divert up to 20 percent of in-laboratory comprehensive respiratory panels. NAAT for flu is reimbursed at $150 and antigen detection tests at $12 or $13. “So you’re going to make tons of money” is what the manufacturers claim, he said. “For our location, it’s really not about reimbursement because at least for Medicare patients, if it’s a hospital-affiliated clinic, there’s no fee for service. If it’s an inpatient, there’s no fee for service. So it does come down to cost versus reimbursement.”
“As much as I would like to move forward with widespread application of POC nucleic acid amplification tests for influenza, there is little robust clinical outcome or health care economics data,” Dr. Nolte said. “It’s developing, and some of the studies look pretty good.” Even so, “there is still a silo mentality in most hospitals with respect to the budget.” If he were to switch MUSC’s POC testing from antigen detection for influenza A and B to NAAT detection, he said, the reagent budget would increase from $100,000 to $450,000. “That’s not an insignificant investment.”
Despite the fun they had debating, Drs. Nolte and Ledeboer ended on a note of full agreement: “We live in a matrixed world.”
And “in a matrixed world,” Dr. Ledeboer said, “that means I might be called to interpret a chemistry test or to interpret a molecular test. And we have to realize that to a certain extent, we are moving toward a more consolidated laboratory. That’s a reality we’re all going to face.” And “it takes a whole village,” he said, “to interpret a lot of these tests.”
Amy Carpenter Aquino is CAP TODAY senior editor.