Sherrie Rice
June 2021—Testing for viral infections post-transplant is an important part of care for transplant patients because of the risk for infections during immunosuppression. But viral load monitoring suffers from interlaboratory variability for several reasons, and while the problem is greater at higher viral loads, the interlaboratory variability is also present at lower viral loads.
Steve Miller, MD, PhD, of the University of California San Francisco, and Joseph Yao, MD, of Mayo Clinic in Rochester, reported the monitoring difficulties, solutions, and new directions last fall in a CAP TODAY webinar made possible by a special educational grant from Roche.
“The quantitation even in the same sample can be quite variable, and that is one of the things we’re trying to deal with,” said Dr. Miller, director of the UCSF clinical microbiology laboratory.
Dr. Yao, a clinical virologist in Mayo’s Division of Clinical Microbiology, shared a case that illustrates the problem.
The patient was a 32-year-old man with end-stage lung disease due to cystic fibrosis, who underwent bilateral lung transplantation three months earlier with a CMV IgG serology positive in the organ donor and recipient. The EBV viral capsid antigen antibody IgG was positive only in the donor but negative in the recipient. The patient presented to his local hospital with two weeks of fever, malaise, night sweats, decreased appetite, and nausea. He was found to be mildly febrile with some jaundice with icteric sclera. He had enlarged palpable lymph nodes in the cervical region of the neck, with enlarged tonsils, and there was a mild tenderness in the right upper quadrant of his abdomen.
His laboratory tests showed mild anemia, a white blood cell count of 4,100 cells/mm3, platelet count 150,000, elevated alanine aminotransferase to 105 U/mL, and elevated total and direct bilirubin (6.0 mg/dL, 4.5 mg/dL). The local hospital was able to obtain the CMV DNA in plasma of 55 IU/mL. They also had a send-out test result obtained for EBV DNA plasma by an external laboratory with 5,500 copies/mL (3.74 log10 copies/mL).
The patient was transferred to a local transplant medical center for further evaluation because of the send-out EBV test result. The transplant center’s EBV DNA viral load test had a result of 85,500 copies/mL (4.93 log10 copies/mL), and this transplant center lab used a laboratory-developed assay with a commercial analyte-specific reagent.
The transplant service team was puzzled by the slightly more than 10-fold difference in the EBV viral load results between the local hospital where the test was sent out and the local transplant medical center lab.
A sample was sent to Mayo Clinic Laboratories, which used a laboratory-developed assay with commercial ASR calibrated to secondary EBV DNA standards (assay range: 100 – 5 × 106 IU/mL, 4.88 log10 IU/mL). The result differed from the result at the original local hospital; it was closer to the copies per mL obtained at the transplant center. The patient had undergone an abdominal CT scan, which showed hypodense lesions in the liver, consistent with and suggestive of post-transplant lymphoproliferative disease.
This case was determined to be a primary EBV infection transmitted by the donor organ leading to PTLD. The patient underwent lymph node biopsy; it was found to be a polymorphic PTLD with early malignant transformation (30 percent) clonal cytogenetic abnormalities and showing immunoglobulin gene rearrangements. He underwent reduction in immunosuppressive therapy, along with B-cell depletion therapy with rituximab, with resolution of post-transplant lymphoproliferative disease.
[dropcap]T[/dropcap]he viruses that can infect transplantation patients are varied, and that’s the first challenge in the lab, Dr. Miller said: “simply having the appropriate assays for them.” Many are human herpesviruses, such as cytomegalovirus, Epstein-Barr virus, human herpesvirus 6, as well as HSV1 and 2. But there are also other categories, such as adenoviruses and polyomaviruses such as BKV. “So you have to have a fairly substantial menu in your lab to do this, and of course each assay has to be validated, perform acceptably, and be monitored through time to make sure the standardization and quantitation are accurate.”
There are few treatment options for these patients, “so often you’re relying on the host immune system and the patient’s own cellular response to be able to clear infection.”
In many cases, the immunosuppressive medications are dialed up or down, “and here you’re walking a tightrope in terms of balancing,” he said, between too much immunosuppression and too little. The lab results are used to help assess the patient. “And this is a difficult thing to do,” Dr. Miller said, “because oftentimes these viruses present nonspecifically.”
“The lab result does impact the management decision, and the change in the viral load is a particularly important parameter to be able to figure out what to do next with your patient.”
Patient management guidelines that address viral load testing and could be applied universally throughout the transplant population would be optimal, he said, but it’s not possible to write them because of the variety of types of transplants and individual patient risk factors. The case of the patient whose CMV donor and recipient status are both seropositive is one in which the risk for that patient to develop CMV changes, for example. “So there are a lot of different factors that need to be taken into account and that makes it difficult to establish a single approach or guidelines for how to do this. It’s an area of active investigation,” Dr. Miller said.
There are technical challenges, too, in developing standard and harmonized assays for virus monitoring. The calibration material is key, he said, and there are a variety of sources and types. “Each assay may be calibrated or standardized to different material. It could be whole virus, a partial genome, or nucleic acid fragments. It may be encapsidated or non-encapsidated, and that makes a big difference during the extraction process. And the efficiency of extraction can affect the overall viral load you achieve at the end.” These materials will perform differently across different assays, he added. “So it is a challenge to find a single calibrator material that will work across all different assay types.”
All of this adds up to interlaboratory variability, Dr. Miller said, and in general, a single lab performing precision studies on the same sample will do well over time. But there’s a two- to four-log variation between laboratories.
Dr. Miller presented a case study of a BK virus infection. The pediatric patient had Wiskott-Aldrich syndrome, an X-linked recessive immune deficiency, and thus received a stem cell transplant for immune system reconstitution. However, the patient had several complications post-transplant: prolonged cytopenias, leading to an elevated risk for opportunistic infections; autoimmune hemolytic anemia, so there was concern about rejection; Salmonella enteritis; low-level CMV viremia, which may or may not be clinically significant but is worrisome because it could suddenly become a severe CMV infection; and hemorrhagic cystitis due to BK virus.
The patient was placed on cidofovir and leflunomide, which are agents—though not effective ones—that can be used to treat BK virus. The treatment was intermittent because there was bone marrow suppression. “As the patient develops the hemorrhagic cystitis, their plasma BK viral level rose substantially, from negative to four or five logs up to over seven logs, and then the patient transferred care. And a second result after five months showed about six and a half logs of virus in the plasma, which is a high level and concerning,” Dr. Miller said.
What is the clinical significance of this latest BK viral load test? he asked. Is it a titer that remains highly elevated and the right step is to reduce immunosuppression and hope the patient is able to combat the virus with their own cellular immunity? Or is the BK virus titer stable and current management should continue? Is another assay needed—BK titer in a urine sample? The fourth option, he said, is to repeat the assay because what’s going on is unknown.
The patient was being seen at the initial transplant center, so UCSF arranged for samples to be sent to its laboratory. “We were able to get a second sample, and that showed what looked like a substantial decline in the viral titer, about one and a half logs, and then in the following month it continued to go down, so the patient was on the right track.” To know whether immune suppression needs to be dialed up or down, he said, “you’re going to be looking at the prevalence of immune-mediated activation such as the hemolytic anemia. Is that getting worse? If it’s not getting worse, you might want to consider staying on.”
The laboratory didn’t decide in this case to assess BK titer in urine. “I look at plasma as being a more specific assay and a better assay to assess the state of the patient,” Dr. Miller said, “so I generally don’t recommend urine testing for assessment of patients with known infection.” Plasma generally will work better, but he recommends that they test on the same assay, if possible, especially if the patient is having substantial complications “and you want to know if it’s going up or down.”
“So this case illustrates the issues we have between the variability between labs and how standardization, if we were able to achieve it, might help in our management so that we wouldn’t have to send samples between labs, with the patients waiting and wondering what’s going on.”
[dropcap]A[/dropcap] WHO reference material has been developed for BK virus. It consists of whole virus preparations of BK virus cell cultured material, which were lyophilized and sent to a variety of laboratories for quantitation; the one that was chosen was a genotype 1b-II, which is predominant in Europe and North America. “Not all patients are infected with the same strain of virus, so you need to consider the material itself—which target is chosen—and the assay,” Dr. Miller said.
These laboratories performed a variety of assays on this standard material, and while they harmonize reasonably well around 7.2 log IU/mL, there is variability depending on the region chosen for targeting with the assay. “That will affect, to some extent, the quantitation, whether you use a target on your assay that is the same as or different from what was used to standardize.”
The goal is to make this universal, but that is often difficult to achieve, Dr. Miller said.
One of the reasons is that a primer mismatch across the genome will affect the amplification efficiency. “Compared to zero primer mismatches, which is the reference of 100 percent efficiency, if you have one primer mismatch, you get around a 10 percent amplification efficiency, rising as you get to low viral loads, because it becomes a little less quantitative and more of a qualitative assessment. That’s a big difference.” With two mutants under their primers, the efficiency is one percent or less. “This is a significant effect and is going to be a major contributor to the difference in viral load if you happen to have primers and your patient has a virus that is mutated under that primer,” Dr. Miller said.
How much does standardization help? In a comparison of the same assay across multiple labs, he said, citing a study that looked at EBV viral load in whole blood measured across 12 labs, “if you standardize to use a single calibrator for these assays, you get a little less variability between labs” (Semenova T, et al. J Clin Microbiol. 2016;54[7]:1746–1750). It doesn’t disappear, he added, but the results between labs correlate better. “So this does have an effect, and it can increase the confidence that your result is meaningful.”
Changes in sample type and processing changes can affect the quantitation. The BK virus standard that he referred to previously was put into various clinical matrices—whole blood, plasma, and urine. “The quantitation changed between each of these different sample types, with the urine having the lowest value assigned after the assay and plasma having the higher value, almost a log higher. While there was tight correlation if you assayed the same sample type, the precision of these assays is good but the matrix effect is substantial.”
The reason for this is likely related to the nucleic acid extraction process, he said, the efficiency of which will be different for different sample types. Plasma might have a higher extraction efficiency than whole blood or urine, which are more complex and have more different materials in them. “And the precision of automated extractions is better than manual extractions, so we can do little things in the lab, like changing to automated systems, that will help reduce the variability in that assay,” Dr. Miller said.
He presented a case study of Epstein-Barr virus in which an adult patient with acute myeloid leukemia received a stem cell transplant following marrow ablation and then three months post-transplant developed fevers. An extensive infectious workup was done, all negative. Paraspinal masses were seen on imaging, so there was concern for post-transplant lymphoproliferative disease. This mass was biopsied by fine-needle aspirate, and the pathology shows only necrotic cell debris. The diagnosis is inconclusive. The plasma EBV PCR was negative. Dr. Miller asked: “Is there a role for whole blood EBV PCR in this case?”
“This gets to the sample type issue, and this is important for Epstein-Barr virus, where you have often relatively low viral loads detectable in patients even with PTLD. And, in general, I see whole blood as being a more sensitive matrix, where you’re assessing both the cellular and the cell-free plasma compartment, and you’re able to detect cell-associated virus.” Plasma is more specific for disease, however, and while the levels in plasma are often lower than in whole blood, he said, detection in plasma is a better indicator that this patient may have PTLD or a significant infection.
“If you compare individual assays for sensitivity and specificity of the diagnosis of PTLD, you see that the specificity in plasma is high, whereas in whole blood, the cellular EBV portion, it’s running in the 80s, maybe up to 90 percent.”
That’s fairly problematic if it’s used as a screening assay, he said, which is often the case in transplant patients, who are monitored over time. “If you have an assay with a low specificity, that’s going to be a problem because you’re going to potentially diagnose many of these cases. If the providers understand this and follow up a whole blood with a plasma assay and assess both, that can be fine.” Whole blood tends to be more sensitive here, and the sensitivity of this assay for detecting PTLD is in the 60 to 70 percent range, sometimes even lower in plasma, Dr. Miller said, “but it can be improved by designing and selecting the right assay for it.”
The approach the UCSF laboratory has taken is to offer plasma EBV viral load testing as more specific. “But in a patient like this who has a high index of suspicion, I would want to do a whole blood EBV PCR. We don’t do that standard in our lab but can send out; it’s something we would do in this case. And in this case it was positive and can help them make the diagnosis and manage the patient.”
What are the solutions to doing this in the laboratory? Assay harmonization can reduce interlaboratory variability. Transitioning to WHO-calibrated standard material can be useful, and here the lab needs to calibrate its standard curve for its laboratory-developed assay if that’s what it’s using. If the lab is using a manufacturer-developed assay, those typically are already calibrated.
The laboratory will want to consider its assay workflow and process. If it can automate steps, that will help improve precision and reduce variability. And the lab will want to develop sample type protocols. “You want to assess, how is this going to be used for your different transplant populations? There’s often a tradeoff between sensitivity and specificity for different sample types,” Dr. Miller said, “and you need to engage with the clinical services on this to make sure they understand how you’re approaching this in the lab and that it does impact how they think about these test results for their patients.”
In some cases, it can be hard to validate every virus for every sample type, and the laboratory may need to send tests out.
What does the future look like for this field? One direction is calibration using standard independent quantitation, and this is with Droplet Digital PCR, in which an absolute copy number is obtained based on the volume assessed. This is beginning to be used to develop reference material, Dr. Miller said. Most labs are going to use commercial reference material, which he said is appropriate. But having an absolute concentration rather than a relative arbitrary concentration level will help to standardize these assays, he said, adding, “The extraction portion is still a major issue for variability.”
Metagenomic next-generation sequencing is a possible mechanism for monitoring transplant patients. The library is prepared typically from plasma cell-free DNA. “The number of sequence reads can be proportional to the viral load, and if you add an internal calibrator or internal control at a known concentration, you can get a relative level from your assay. And there are novel approaches to this,” Dr. Miller said.
Finally, fully automated assays that are precalibrated are important, he said, but they still need validation of alternative sample types.
[dropcap]D[/dropcap]r. Yao defines commutability as the ability of a reference or control material for measurements of an analyte to have interassay properties comparable to the properties of authentic clinical samples when tested by more than one assay method. This material is considered to be commutable, he said, when an assay produces the same result for this material as it does for an authentic patient sample that contained the same analyte concentration. Assays calibrated with a commutable reference material will produce results for clinical samples that are equivalent among all such assays. That means the results are traceable to the reference system and there is no calibration bias among the assays when the commutable reference material is used. The current WHO standard is not foolproof, he said, and there’s still work to do to improve the commutability of the different reference materials for different assays currently in clinical use.
In the June 2020 CAP Viral Load Survey, all 298 participant laboratories were officially reporting results for CMV in IU/mL. Almost 90 percent were still using ASR for CMV; 30 laboratories were using LDT. “For EBV, it’s half and half,” Dr. Yao reported. Half were reporting in copies/mL and half in IU/mL. “For BK virus it’s the other way around. It’s a 30/70 split,” he said. Adenovirus currently has an international standard, but for the 60 laboratory participants, they’re only reporting in copies/mL. HHV-6 also has a WHO international standard, and no one was reporting in IU/mL at that time.
To improve post-transplantation viral load monitoring, Dr. Yao recommends the following:
- Use commutable reference and control materials that are traceable to WHO international standards (IU) for assay calibration. WHO standards are available for adenovirus, BKV, EBV, HHV-6, and CMV.
- Use quantification assays that are calibrated to WHO IU.
- Even when assays are calibrated to international standards and using commutable reference materials or controls, because of variability from one assay to another, try to use the same assay to monitor viral load results over time in any given patient, especially if commutability of assay calibration material is unknown.
Sherrie Rice is editor of CAP TODAY.