|Values are valid only on day of printing.|
Click CC to turn on closed captioning.
Published: March 2010Print Record of Viewing
Dr. Grys discusses the utility of assays used to quantify Cytomegalovirus and Epstein-Barr virus. He also describes the variability inherent to measuring a viral load, and how this must be considered when interpreting results in a clinical context.
Presenter: Thomas E. Grys, PhD
Welcome to Mayo Medical Laboratories' Hot Topics. These presentations provide short discussion of current topics and may be helpful to you in your practice. Our presenter for this program is Dr. Thomas Grys, Associate Laboratory Director, Mayo Medical Laboratories New England, Department of Laboratory Medicine and Pathology at Mayo Clinic. Dr. Grys will discuss the utility of assays used to quantify Cytomegalovirus and Epstein-Barr virus. He will also describe the variability inherent to measuring a viral load, and how this must be considered when interpreting results in a clinical context.
In this hot topic, we will discuss assays used to quantify 2 viruses, Cytomegalovirus and Epstein-Barr virus. This is an area of active research, and there are a lot of details that affect both the measurement and interpretation of viral loads.
The 2 viruses I will be discussing today are both in the Herpesviridae family.
Cytomegalovirus, or CMV, is a virus that infects a majority of the population, and by adulthood, most people are seropositive for the virus, meaning that they have been infected by it in the past. The virus can infect many cell types, but usually targets lymphocytes and epithelial cells. This virus establishes and maintains a latent infection that can reactivate later in life. The disease manifestations for CMV are myriad, and include mononucleosis-like symptoms as well as tissue-specific disease, such as retinitis, pneumonia, and colitis.
Epstein-Barr Virus, or EBV, primarily infects B cells, and, like CMV, causes a latent infection that can also reactivate later in life. The vast majority of the population has been infected by this virus by adulthood, and so a large percentage are seropositive for this virus. The more common presentation for EBV infection is mononucleosis, but it can also cause Burkitt's lymphoma and nasopharyngeal carcinoma, though these are far less common.
The context for our discussion is primarily transplant patients who, because of their state of immunosuppression, have an increased susceptibility to infections in general. I will be focusing the discussion today on CMV and EBV.
In broad terms, when considering the total risk of EBV or CMV infection, one must consider the net state of immunosuppression, including any immunosuppression drugs, antiviral drugs, and any cell growth drugs that the patient may be taking.
Transplant patients can be grouped into 2 main categories: recipients of solid organs and recipients of hematopoetic stem cells. It is important to understand that there are many details providing important points of differentiation within each group, but these considerations are outside the scope of this discussion.
Patients receiving a solid organ are most at risk for CMV in the first few months following the transplant, or following the end of antiviral prophylaxis. In fact, a substantial percentage of transplant patients would develop symptomatic disease if prophylactic or preemptive therapy were not employed. The most common manifestations for CMV disease in this population is gasteroenteritis and retinitis, but CMV syndrome is also common, resulting in nonspecific symptoms such as fever, weakness, myalgia, and myelosuppression.
Stem cell transplant patients will often present with CMV syndrome or pneumonia, and are also at highest risk within the first few months following transplant.
Although it is thought that some cases of tissue-invasive disease can produce high viral loads, it is important to bear in mind that viral loads may be negative for infections that are localized to a single tissue, or to the transplanted organ. Therefore, a negative viral load does not preclude the diagnosis of tissue-invasive disease. Symptoms consistent with localized CMV infection, combined with a negative viral load may be reason to pursue alternative strategies, such as a biopsy.
The decision to treat CMV must be informed by a number of factors: The patient and transplant type, the laboratory method used to quantify the viral load, as well as the specimen submitted to the laboratory.
In the literature, there are many proposed decision points for treatment of CMV, anywhere from 200 (or 2.3 log) copies per milliliter to 86,000 (or 4.9 log) copies per milliliter. The range of values is dependent not only on differences in the patient population, but also in the different methodologies used to determine the viral loads
One approach is to stratify patients by risk type. In this model, high-risk patients should have the lowest threshold, or tolerance before treatment, whereas moderate or lower-risk patients may not require treatment unless the viral load passes a higher threshold. Please note that these numbers are examples, and may not apply to your local testing methodology.
For patients at risk of infection, there are 2 main strategies to prevent CMV disease.
The first is prophylactic. In this strategy, high-risk patients are given treatment during the period of time in which they are thought to be most at risk for disease. The treatment is given regardless of symptoms, and is intended to prevent primary infection or reactivation. The advantage is that the patient is being continuously protected from CMV infections, but the disadvantage is that the drugs can be costly and are not without drug-drug interactions and side effects, not to mention the increased potential for antiviral resistance.
The second strategy is preemptive. For some low-risk or moderate-risk patients, this might be the preferred strategy. Viral loads are monitored periodically, typically weekly. Therapy is then administered when CMV is first detected, regardless of whether symptoms are evident. The advantage here is that only infections are being treated, which reduces drug complications and reduces the potential for antiviral resistance. The disadvantage is that the viral load tests are not without cost, and the therapy may not begin as soon as needed.
Treatment for CMV in transplant populations is typically ganciclovir, though a reduction in immunosuppression may also be considered.
Transplant patients with EBV infections are at risk for post-transplantation lymphoproliferative disorder, or PTLD, which can manifest anywhere along a spectrum of disease from mononucleosis-like symptoms to lymphoma.
PTLDs are caused by an abnormal proliferation of EBV-infected B cells, which result from the lack of adequate T-cell function that usually keep B-cell replication under control. Treatments or conditions that reduce T-cell function in the patient increase the risk for PTLD.
Although the mortality can be quite high, the incidence of PTLD is fairly low, and depends on factors such as the organ transplanted, age and seropositivity of the patient, and the type of immunosuppressive regimen.
Transplant patients are most at risk for EBV-associated PTLD within the first few months after transplant, although PTLD can also occur months or years following transplantation. Many different specimen types have been evaluated for use in quantitative EBV assays, including whole blood, plasma, and various cell fractions.
The most commonly used specimen is whole blood.
Quantitative EBV assays are sensitive, but not entirely specific for PTLD disease. That is to say that most patients who develop PTLD show increasing or high levels of EBV virus before symptom onset, though some patients with a high level of virus may remain asymptomatic. A negative result by a PCR assay for EBV has a high negative predictive value for PTLD, meaning that the short-term risk for PTLD is very low.
In the literature, proposed decision points for treatment are reported anywhere from 1000 to 50,000 (3 log to 4.7 log) copies/mL. Like CMV, factors that contribute to this broad range of numbers include patient populations, patient therapies, and laboratory detection methods.
Treatments for EBV-infected patients are usually targeted towards the abnormal B-cell proliferation. Like CMV, a preemptive strategy has been proposed to be a way to balance the risks and benefits of treatment modalities.
The first treatment decision is often to reduce immunosuppression of the patient to allow the body's own defenses to control the B-cell replication. Another option is to treat with Rituximab, which specifically targets B cells. There are other treatments being studied, such as cytotic T-cell transfer, wherein a patient is infused with a specific immune cell type that helps to control B-cell replication.
Unfortunately the risks of treatment are not benign, and include not only drug-drug interactions or side effects, but also could result in a rejection of the transplanted organ for solid organ transplant recipients, or graft-versus-host disease in the case of stem cell transplant recipients.
To put what weve discussed so far into perspective, let's imagine that a physician is presented with a patient whose symptoms are fever and malaise. This nonspecific presentation could indicate an infection, or it could be a sign of rejection. The treatment decisions for these 2 possibilities may be in total opposition. The wrong decision could exacerbate the real problem and have severe consequences.
If the patient is thought to have an infection, the reduction in immunosuppression can strengthen the body's ability to fight the infection, but it also increases the risk of rejection or graft vs host disease.
Alternatively, if the patient is thought to be showing signs of rejection, increasing immunosuppression could save the transplant, but allow an infection or PTLD to develop.
An added difficulty is that whatever decision is made, the effects will not be fully known for several days, up to a week. This is because the existing drugs will take time to be cleared from the body, and any new treatments or dosages may take several days to achieve therapeutic levels.
Quantitative viral load assays can provide information to help make these difficult treatment decisions. Common questions regarding these tests include: What is the threshold between insignificant viral replication and risk for disease? What is the interval of change that is significant?
To better understand viral load assays, we must understand how they work. In the lab, we use a particular method and employ known viral standards to calculate the amount of virus in a patient specimen and report values in appropriate units. As we advance to the next slide, I will highlight some of the variables at play in this process.
As you can see, there are a lot of points of variability. In the following slides, I will discuss how these variables can affect the results of viral load assays, and what the potential implications are for patient care.
An assay must first be calibrated with a known amount of virus or nucleic acid before it can be used to determine the level of virus in a patient specimen. The types of calibrators that are commonly used range from live virus to dead virus to viral nucleic acid. The matrix, or solution, in which the calibrator material is suspended can also affect the performance of the assay. After calibration, there are other variables to consider, such as the technology used for viral quantification,the specimen source being analyzed, and the units in which the results are reported.
All of these variables mean that results of viral load assays are usually not comparable between laboratories, and therefore, decision points must be determined locally for each practice.
Ideally, there would be an international standard material consisting of a known amount of virus, which would then serve as a reference material as laboratories and manufacturers design their assays. In this way, the results from all laboratories would be traceable to the same gold standard. International standards for HIV, hepatitis C, and hepatitis B are already available.
Although there are groups working on standards for CMV & EBV, no international standards exist currently, and therefore all assays are only relative to themselves.
Once the laboratory produces the result, the physician needs to interpret what the result means. To begin this discussion, I would like you to consider the following number comparisons. Is 10 different than 15? 15 is 5 units higher 1.5 times higher Is 10 very different than 100? 100 is 90 units 10 times higher
If these were dollar amounts, I would certainly choose 100, wouldn't you? Now, if the choice was between 500,000 and 500,005 dollars, I don't think I would mind which amount I received. Although this is like the first comparison, where the numbers were 5 units apart, the higher value is now only 1.00001 times higher.
In the final example, 500,000 and 500,090 they are 90 units apart, just like the 2nd example. However, whereas the numbers in the 2nd example seemed vastly different, these two numbers seem roughly equivalent, since the 500,090 is only 1.00018 times higher.
We will return to these numbers later, but the point thus far is that the value inherent to a result depends in large part on the range of possible values.
During an active infection, each time a virus infects a cell, it may produce 10s or 100s of viruses. Then, each of these viruses infects other cells and produces 10s or 100s more viruses. This is an example of exponential growth, which is best measured on a logarithmic scale.
Similarly, in the laboratory, most viral load assays use polymerase chain reaction, or PCR. PCR typically uses 25-50 cycles of amplification. Each cycle doubles the amount of a target sequence of nucleic acid. This is also a logarithmic process.
Since both the biological phenomenon and the laboratory method measuring it grow logarithmically, it makes sense to think about viral loads in log terms.
The advantages of log do not stop at the better approximation of biological and laboratory phenomena. The transformation of integer values to log means that the results within a population are more closely approximate to a normal, or bell-curved, distribution. This is the first assumption necessary when using traditional statistics, and sound statistics are vital in providing evidence-based criteria for run acceptance in the laboratory.
Log numbers should also allow physicians to more easily trend patient viral loads. For example, if a patient viral load changed from 2000 to 4000 copies/mL (or 3.3 log to 3.6 log), the precision data from the literature and from our laboratory tells us that these numbers are statistically equivalent. Although the viral load has doubled, the reality is that because of biological and laboratory variation, there has been no significant change in viral load. On the log scale, the change was only 0.3 log, which is much less alarming.
Now let's take a moment to review integer and log numbers.
In the table shown here, each row shows the same value in 3 different numbering schemes: integer, scientific notation, and log-based 10. The first column is integers, which we know and love. We grew up with 1, 2, 3, 4, 5, etc. In the middle column is scientific notation. This is a short-hand way to save writing a lot of digits when a large number is involved. So, 1000 becomes 1.0 x 10 to the third power. This lets us know to put 3 zeroes behind the 1. In the log scale, 1000 is 3. You may notice that the exponent 3 from scientific notation is helpful when thinking in log. This is often helpful to some extent. The second row, one million, has 6 zeroes, so scientific notation is 1.0 x 10 to the sixth power. In log with a base of 10, the value is 6.
Now let's look at the blue text below the table. Log numbers use different "bases," the most common of which is ten. What this means is that the value of the log, in this case, 3, is used as the exponent for the base. In other words, 10 to the third power is 1000. We can write this out as 10 times 10 times 10.
In the second example, the base is 2. If we raise 2 to the power of 3, we have 2 times 2 times 2, equals 8.
In the third example, we would raise the base 10 to the 3.74 power. This one is harder to write out, but the result is that the log of the integer 5,500 is 3.74. In the last row of the table, you can see how this value differs from that of scientific notation.
So how do changes on the integer scale compare to changes on the log scale? In this table, I demonstrate 3 pairs of numbers. The black text is integer values, and the blue text is log values. The middle column shows the fold difference between the integer values. For example, in the first row, 100 is 10 fold higher than 10. The column on the right shows the difference of the log numbers. The log of 100 is 2, and the log of 10 is 1. Thus, the difference between those values is 1. In the middle row, you can see that a 3 fold change is about half a log difference. This is a helpful relationship to remember. Finally, in the third row, we can see that a 2-fold difference, in this case 20,000 to 40,000 is a difference of about 0.3 log.
Now let's touch on the potential disadvantages of log numbers. The biggest hurdle is that laboratory technologists and physicians are often unfamiliar with log concepts and are not comfortable using log numbers.
Another criticism is that it is hard to calculate log in your head, or to compare numbers. But, let's look at the numbers below. Since we know that the number of zeroes after 1 can help us to estimate log numbers, we can use that to stake out a roadmap. Then, if we remember just one other relationship, we can begin to navigate the world of log numbers.
At each midpoint between these numbers, 5000, 50,000, and 500,000, we can see that the log value has a decimal of 0.7. Using these concepts, we can already estimate that 8000 has a log value of more than 3.7, but less than 4. A value of 400,000 has a log value of more than 5, but less than 5.7.
Remember this exercise? If we convert the pairs of values to the log scale, we can see that what seemed to make intuitive sense is supported by the log numbers: If these were viral loads, and we used the 0.5 log interval as a marker of significant change, then we can see that 10 is different than 100, but 500,000 and 500,090 are essentially indistinguishable.
Now let's return again to the issue of variability. Two recent studies sent the same samples to 30 laboratories and only about half of the results agreed within 0.5 log (remember, this is a 3-fold change!). This lack of precision between laboratories is not unexpected, since different labs are calibrating in different ways and using different assays.
What about precision within a laboratory? Let us imagine that a laboratory can test the same specimen over and over with a level of precision where the standard deviation is about 0.2 log. In statistics, 2 standard deviations encompass roughly 95% of the population. In other words, testing the same sample multiple times would produce values that are generally no greater than 0.4 log from the mean.
To put this variability into perspective, let's think about the significant change in HIV. For HIV, we are looking for an initial 1 log drop in response to treatment, and generally changes of less than 0.5 log are not considered to be not clinically significant, or different, than a previous value. If we borrow the same 0.5 log concept for CMV and EBV, then the laboratory needs to be producing results that have less variability than 0.5 log. In the example here, this means that the controls should be running within a range of 2 standard deviations, otherwise, the run is not valid. If a laboratory has an assay with an unacceptable level of precision, then the results may not be reflective of true changes in patient viral loads.
What level of change is significant, or "real," in viral load assays? It is thought that variation within the patient is about 0.2 or 0.3 log. This means that 2 samples drawn at different times on the same day could vary by 2-fold. Within the laboratory, as we just discussed on the previous slide, the variation is another 0.2 or 0.3 log.
As you can see, a patient's viral load must change substantially to be outside the variability inherent in the biological and laboratory processes.
Let's pause for a moment to compare the state of the art for viral load testing in HIV and Hepatitis C versus CMV and EBV. You can see that for HIV and HCV, there are defined reporting units, international standard reference materials, and defined treatment goals. It is also generally regarded that in order for a viral load to be significantly different than the previous one, HIV should change by about 0.5 log or more (at least 3 fold), whereas HCV should change by about 1 log (or 10 fold). Any differences less than that are to be expected because of the variability of the viremia in the patient and the variation of the laboratory assays.
When tracking viral loads in patients with HIV or hepatitis C, the treatment goals are becoming well-established. Initiation of effective treatment for HIV should result in decreasing viral loads at a rate of 1 log per month (or 90% lower than each previous month). For HCV, demonstration of a 2 log, or 100 fold drop in viral load, in the first 12 weeks is an indication that the patient stands a good chance of successful treatment in the long term.
For CMV & EBV, there are a multitude of assays and reporting units, no international reference material, no defined treatment goals and very loose concepts of significant change.
There are a lot of interesting studies on CMV and EBV viral load testing in the literature. The problem is determining how best to apply the information to your local practice. It is essential to remember that any decision points or treatment guidelines that are published by various groups are HIGHLY dependent on the patient population, specimen analyzed, methodology, calibration used, and reporting units.
Although we cannot necessarily apply exact numbers or decision point levels from other studies, the literature can certainly be helpful in determining what concepts to consider when interpreting viral loads.
The graph on the left is a theoretical example of 2 patients' viral load profiles plotted over time. Although we do not see any numbers on the Y axis, we can still make some interpretations from the trends that these viral loads show. Points to consider are the level of the first positive specimen, the highest viral load level detected, the rate of increase or decrease, and whether the patient is positive (and at what level) at the end of treatment. These variables, along with clinical presentation and other test results, can help a physician to determine: whether the patient might require a drug dosage adjustment, whether treatment should be extended, whether the virus might have become resistant to the medication, and estimate the risk recurrence of disease when the treatment ends.
To summarize: Viral load assays are an important tool in successful management of transplant patients. Treatment strategies can include prophylaxis, or monitoring with preemptive therapy. Although test results, and therefore decision points, are not comparable among laboratories, concepts relating to viral load changes may be transferable. In general, lower viral loads indicate a lower risk of disease, whereas a trend of increasing or high viral loads may indicate a higher risk for disease. Finally, as with most treatment modalities that rely heavily on the laboratory, good communication between care providers and laboratorians is essential.